从BaseTool 中拆分除GPU的Base、RTPC、TBPImage 等GPU文件
parent
16ed299e9f
commit
7ad0acdd5d
|
|
@ -1,953 +0,0 @@
|
|||
|
||||
|
||||
#include <iostream>
|
||||
#include <memory>
|
||||
#include <cmath>
|
||||
#include <complex>
|
||||
#include <device_launch_parameters.h>
|
||||
#include <cuda_runtime.h>
|
||||
#include <cublas_v2.h>
|
||||
#include <cuComplex.h>
|
||||
|
||||
#include "BaseConstVariable.h"
|
||||
#include "GPUTool.cuh"
|
||||
|
||||
#ifdef __CUDANVCC___
|
||||
|
||||
#define CUDAMEMORY Memory1MB*100
|
||||
|
||||
#define LAMP_CUDA_PI 3.141592653589793238462643383279
|
||||
|
||||
|
||||
// 定义参数
|
||||
__device__ cuComplex cuCexpf(cuComplex x)
|
||||
{
|
||||
float factor = exp(x.x);
|
||||
return make_cuComplex(factor * cos(x.y), factor * sin(x.y));
|
||||
}
|
||||
|
||||
// 定义仿真所需参数
|
||||
|
||||
__device__ float GPU_getSigma0dB(CUDASigmaParam param, float theta) {
|
||||
float sigma= param.p1 + param.p2 * exp(-param.p3 * theta) + param.p4 * cos(param.p5 * theta + param.p6);
|
||||
return sigma;
|
||||
}
|
||||
|
||||
__device__ CUDAVector GPU_VectorAB(CUDAVector A, CUDAVector B) {
|
||||
CUDAVector C;
|
||||
C.x = B.x - A.x;
|
||||
C.y = B.y - A.y;
|
||||
C.z = B.z - A.z;
|
||||
return C;
|
||||
}
|
||||
|
||||
__device__ float GPU_VectorNorm2(CUDAVector A) {
|
||||
return sqrtf(A.x * A.x + A.y * A.y + A.z * A.z);
|
||||
}
|
||||
|
||||
__device__ float GPU_dotVector(CUDAVector A, CUDAVector B) {
|
||||
return A.x * B.x + A.y * B.y + A.z * B.z;
|
||||
}
|
||||
|
||||
__device__ float GPU_CosAngle_VectorA_VectorB(CUDAVector A, CUDAVector B) {
|
||||
return GPU_dotVector(A, B) / (GPU_VectorNorm2(A) * GPU_VectorNorm2(B));
|
||||
}
|
||||
|
||||
__device__ CUDAVectorEllipsoidal GPU_SatelliteAntDirectNormal(float RstX, float RstY, float RstZ,
|
||||
float antXaxisX, float antXaxisY, float antXaxisZ,
|
||||
float antYaxisX, float antYaxisY, float antYaxisZ,
|
||||
float antZaxisX, float antZaxisY, float antZaxisZ,
|
||||
float antDirectX, float antDirectY, float antDirectZ
|
||||
) {
|
||||
CUDAVectorEllipsoidal result{ 0,0,-1 };
|
||||
float Xst = -1 * RstX; // 卫星 --> 地面
|
||||
float Yst = -1 * RstY;
|
||||
float Zst = -1 * RstZ;
|
||||
float AntXaxisX = antXaxisX;
|
||||
float AntXaxisY = antXaxisY;
|
||||
float AntXaxisZ = antXaxisZ;
|
||||
float AntYaxisX = antYaxisX;
|
||||
float AntYaxisY = antYaxisY;
|
||||
float AntYaxisZ = antYaxisZ;
|
||||
float AntZaxisX = antZaxisX;
|
||||
float AntZaxisY = antZaxisY;
|
||||
float AntZaxisZ = antZaxisZ;
|
||||
// 天线指向在天线坐标系下的值
|
||||
float Xant = (Xst * (AntYaxisY * AntZaxisZ - AntYaxisZ * AntZaxisY) + Xst * (AntXaxisZ * AntZaxisY - AntXaxisY * AntZaxisZ) + Xst * (AntXaxisY * AntYaxisZ - AntXaxisZ * AntYaxisY)) / (AntXaxisX * (AntYaxisY * AntZaxisZ - AntZaxisY * AntYaxisZ) - AntYaxisX * (AntXaxisY * AntZaxisZ - AntXaxisZ * AntZaxisY) + AntZaxisX * (AntXaxisY * AntYaxisZ - AntXaxisZ * AntYaxisY));
|
||||
float Yant = (Yst * (AntYaxisZ * AntZaxisX - AntYaxisX * AntZaxisZ) + Yst * (AntXaxisX * AntZaxisZ - AntXaxisZ * AntZaxisX) + Yst * (AntYaxisX * AntXaxisZ - AntXaxisX * AntYaxisZ)) / (AntXaxisX * (AntYaxisY * AntZaxisZ - AntZaxisY * AntYaxisZ) - AntYaxisX * (AntXaxisY * AntZaxisZ - AntXaxisZ * AntZaxisY) + AntZaxisX * (AntXaxisY * AntYaxisZ - AntXaxisZ * AntYaxisY));
|
||||
float Zant = (Zst * (AntYaxisX * AntZaxisY - AntYaxisY * AntZaxisX) + Zst * (AntXaxisY * AntZaxisX - AntXaxisX * AntZaxisY) + Zst * (AntXaxisX * AntYaxisY - AntYaxisX * AntXaxisY)) / (AntXaxisX * (AntYaxisY * AntZaxisZ - AntZaxisY * AntYaxisZ) - AntYaxisX * (AntXaxisY * AntZaxisZ - AntXaxisZ * AntZaxisY) + AntZaxisX * (AntXaxisY * AntYaxisZ - AntXaxisZ * AntYaxisY));
|
||||
// 计算theta 与 phi
|
||||
float Norm = sqrtf(Xant * Xant + Yant * Yant + Zant * Zant); // 计算 pho
|
||||
float ThetaAnt = acosf(Zant / Norm); // theta 与 Z轴的夹角
|
||||
float YsinTheta = Yant / sinf(ThetaAnt);
|
||||
float PhiAnt = (YsinTheta / abs(YsinTheta)) * acosf(Xant / (Norm * sinf(ThetaAnt)));
|
||||
result.theta = ThetaAnt;
|
||||
result.phi = PhiAnt;
|
||||
result.pho = Norm;
|
||||
return result;
|
||||
}
|
||||
|
||||
/**
|
||||
天线方向图插值方法，以双线性插值算法为基础，由theta与phi组合得到的矩阵图为基础数据，通过插值计算的方法获取目标点的数据。
|
||||
其中行是theta、列是phi
|
||||
*/
|
||||
__device__ float GPU_BillerInterpAntPattern(float* antpattern,
|
||||
float starttheta, float startphi, float dtheta, float dphi,
|
||||
long thetapoints, long phipoints,
|
||||
float searththeta, float searchphi) {
|
||||
float stheta = searththeta;
|
||||
float sphi = searchphi;
|
||||
if (stheta > 90) {
|
||||
return 0;
|
||||
}
|
||||
else {}
|
||||
|
||||
|
||||
float pthetaid = (stheta - starttheta) / dtheta;//
|
||||
float pphiid = (sphi - startphi) / dphi;
|
||||
|
||||
long lasttheta = floorf(pthetaid);
|
||||
long nextTheta = lasttheta + 1;
|
||||
long lastphi = floorf(pphiid);
|
||||
long nextPhi = lastphi + 1;
|
||||
|
||||
|
||||
if (lasttheta < 0 || nextTheta < 0 || lastphi < 0 || nextPhi < 0 ||
|
||||
lasttheta >= thetapoints || nextTheta >= thetapoints || lastphi >= phipoints || nextPhi >= phipoints)
|
||||
{
|
||||
return 0;
|
||||
}
|
||||
else {
|
||||
float x = stheta;
|
||||
float y = sphi;
|
||||
|
||||
float x1 = lasttheta * dtheta + starttheta;
|
||||
float x2 = nextTheta * dtheta + starttheta;
|
||||
float y1 = lastphi * dphi + startphi;
|
||||
float y2 = nextPhi * dphi + startphi;
|
||||
|
||||
float z11 = antpattern[lasttheta * phipoints + lastphi];
|
||||
float z12 = antpattern[lasttheta * phipoints + nextPhi];
|
||||
float z21 = antpattern[nextTheta * phipoints + lastphi];
|
||||
float z22 = antpattern[nextTheta * phipoints + nextPhi];
|
||||
|
||||
|
||||
//z11 = powf(10, z11 / 10); // dB-> 线性
|
||||
//z12 = powf(10, z12 / 10);
|
||||
//z21 = powf(10, z21 / 10);
|
||||
//z22 = powf(10, z22 / 10);
|
||||
|
||||
float GainValue = (z11 * (x2 - x) * (y2 - y)
|
||||
+ z21 * (x - x1) * (y2 - y)
|
||||
+ z12 * (x2 - x) * (y - y1)
|
||||
+ z22 * (x - x1) * (y - y1));
|
||||
GainValue = GainValue / ((x2 - x1) * (y2 - y1));
|
||||
return GainValue;
|
||||
}
|
||||
}
|
||||
|
||||
__device__ cuComplex GPU_calculationEcho(float sigma0, float TransAnt, float ReciveAnt,
|
||||
float localangle, float R, float slopeangle, float Pt, float lamda) {
|
||||
float r = R;
|
||||
float amp = Pt * TransAnt * ReciveAnt;
|
||||
amp = amp * sigma0;
|
||||
amp = amp / (powf(4 * LAMP_CUDA_PI, 2) * powf(r, 4)); // 反射强度
|
||||
float phi = (-4 * LAMP_CUDA_PI / lamda) * r;
|
||||
cuComplex echophi = make_cuComplex(0, phi);
|
||||
cuComplex echophiexp = cuCexpf(echophi);
|
||||
cuComplex echo;
|
||||
echo.x = echophiexp.x * amp;
|
||||
echo.y = echophiexp.y * amp;
|
||||
return echo;
|
||||
}
|
||||
|
||||
__global__ void CUDA_DistanceAB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* R, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
R[idx] = sqrtf(powf(Ax[idx] - Bx[idx], 2) + powf(Ay[idx] - By[idx], 2) + powf(Az[idx] - Bz[idx], 2));
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_B_DistanceA(float* Ax, float* Ay, float* Az, float Bx, float By, float Bz, float* R, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
R[idx] = sqrtf(powf(Ax[idx] - Bx, 2) + powf(Ay[idx] - By, 2) + powf(Az[idx] - Bz, 2));
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_make_VectorA_B(float sX, float sY, float sZ, float* tX, float* tY, float* tZ, float* RstX, float* RstY, float* RstZ, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
RstX[idx] = sX - tX[idx]; // 地面->天
|
||||
RstY[idx] = sY - tY[idx];
|
||||
RstZ[idx] = sZ - tZ[idx];
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_Norm_Vector(float* Vx, float* Vy, float* Vz, float* R, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
R[idx] = sqrtf(powf(Vx[idx], 2) + powf(Vy[idx], 2) + powf(Vz[idx], 2));
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_cosAngle_VA_AB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* anglecos, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
float tAx = Ax[idx];
|
||||
float tAy = Ay[idx];
|
||||
float tAz = Az[idx];
|
||||
float tBx = Bx[idx];
|
||||
float tBy = By[idx];
|
||||
float tBz = Bz[idx];
|
||||
float AR = sqrtf(powf(tAx, 2) + powf(tAy, 2) + powf(tAz, 2));
|
||||
float BR = sqrtf(powf(tBx, 2) + powf(tBy, 2) + powf(tBz, 2));
|
||||
float dotAB = tAx * tBx + tAy * tBy + tAz * tBz;
|
||||
float result = acosf(dotAB / (AR * BR));
|
||||
anglecos[idx] = result;
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_SatelliteAntDirectNormal(float* RstX, float* RstY, float* RstZ,
|
||||
float antXaxisX, float antXaxisY, float antXaxisZ,
|
||||
float antYaxisX, float antYaxisY, float antYaxisZ,
|
||||
float antZaxisX, float antZaxisY, float antZaxisZ,
|
||||
float antDirectX, float antDirectY, float antDirectZ,
|
||||
float* thetaAnt, float* phiAnt
|
||||
, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
float Xst = -1 * RstX[idx]; // 卫星 --> 地面
|
||||
float Yst = -1 * RstY[idx];
|
||||
float Zst = -1 * RstZ[idx];
|
||||
float AntXaxisX = antXaxisX;
|
||||
float AntXaxisY = antXaxisY;
|
||||
float AntXaxisZ = antXaxisZ;
|
||||
float AntYaxisX = antYaxisX;
|
||||
float AntYaxisY = antYaxisY;
|
||||
float AntYaxisZ = antYaxisZ;
|
||||
float AntZaxisX = antZaxisX;
|
||||
float AntZaxisY = antZaxisY;
|
||||
float AntZaxisZ = antZaxisZ;
|
||||
|
||||
// 归一化
|
||||
float RstNorm = sqrtf(Xst * Xst + Yst * Yst + Zst * Zst);
|
||||
float AntXaxisNorm = sqrtf(AntXaxisX * AntXaxisX + AntXaxisY * AntXaxisY + AntXaxisZ * AntXaxisZ);
|
||||
float AntYaxisNorm = sqrtf(AntYaxisX * AntYaxisX + AntYaxisY * AntYaxisY + AntYaxisZ * AntYaxisZ);
|
||||
float AntZaxisNorm = sqrtf(AntZaxisX * AntZaxisX + AntZaxisY * AntZaxisY + AntZaxisZ * AntZaxisZ);
|
||||
|
||||
|
||||
float Rx = Xst / RstNorm;
|
||||
float Ry = Yst / RstNorm;
|
||||
float Rz = Zst / RstNorm;
|
||||
float Xx = AntXaxisX / AntXaxisNorm;
|
||||
float Xy = AntXaxisY / AntXaxisNorm;
|
||||
float Xz = AntXaxisZ / AntXaxisNorm;
|
||||
float Yx = AntYaxisX / AntYaxisNorm;
|
||||
float Yy = AntYaxisY / AntYaxisNorm;
|
||||
float Yz = AntYaxisZ / AntYaxisNorm;
|
||||
float Zx = AntZaxisX / AntZaxisNorm;
|
||||
float Zy = AntZaxisY / AntZaxisNorm;
|
||||
float Zz = AntZaxisZ / AntZaxisNorm;
|
||||
|
||||
float Xant = (Rx * Yy * Zz - Rx * Yz * Zy - Ry * Yx * Zz + Ry * Yz * Zx + Rz * Yx * Zy - Rz * Yy * Zx) / (Xx * Yy * Zz - Xx * Yz * Zy - Xy * Yx * Zz + Xy * Yz * Zx + Xz * Yx * Zy - Xz * Yy * Zx);
|
||||
float Yant = -(Rx * Xy * Zz - Rx * Xz * Zy - Ry * Xx * Zz + Ry * Xz * Zx + Rz * Xx * Zy - Rz * Xy * Zx) / (Xx * Yy * Zz - Xx * Yz * Zy - Xy * Yx * Zz + Xy * Yz * Zx + Xz * Yx * Zy - Xz * Yy * Zx);
|
||||
float Zant = (Rx * Xy * Yz - Rx * Xz * Yy - Ry * Xx * Yz + Ry * Xz * Yx + Rz * Xx * Yy - Rz * Xy * Yx) / (Xx * Yy * Zz - Xx * Yz * Zy - Xy * Yx * Zz + Xy * Yz * Zx + Xz * Yx * Zy - Xz * Yy * Zx);
|
||||
|
||||
|
||||
// 计算theta 与 phi
|
||||
float Norm = sqrtf(Xant * Xant + Yant * Yant + Zant * Zant); // 计算 pho
|
||||
float ThetaAnt = acosf(Zant / Norm); // theta 与 Z轴的夹角
|
||||
float PhiAnt = atanf(Yant / Xant); // -pi/2 ~pi/2
|
||||
|
||||
|
||||
if (abs(Yant) < PRECISIONTOLERANCE) { // X轴上
|
||||
PhiAnt = 0;
|
||||
}
|
||||
else if (abs(Xant) < PRECISIONTOLERANCE) { // Y轴上，原点
|
||||
if (Yant > 0) {
|
||||
PhiAnt = PI / 2;
|
||||
}
|
||||
else {
|
||||
PhiAnt = -PI / 2;
|
||||
}
|
||||
}
|
||||
else if (Xant < 0) {
|
||||
if (Yant > 0) {
|
||||
PhiAnt = PI + PhiAnt;
|
||||
}
|
||||
else {
|
||||
PhiAnt = -PI+PhiAnt ;
|
||||
}
|
||||
}
|
||||
else { // Xant>0 X 正轴
|
||||
|
||||
}
|
||||
|
||||
if (isnan(PhiAnt)) {
|
||||
printf("V=[%f,%f,%f];norm=%f;thetaAnt=%f;phiAnt=%f;\n", Xant, Yant, Zant,Norm, ThetaAnt, PhiAnt);
|
||||
}
|
||||
|
||||
//if (abs(ThetaAnt - 0) < PRECISIONTOLERANCE) {
|
||||
// PhiAnt = 0;
|
||||
//}
|
||||
//else {}
|
||||
|
||||
|
||||
thetaAnt[idx] = ThetaAnt*r2d;
|
||||
phiAnt[idx] = PhiAnt*r2d;
|
||||
//printf("Rst=[%f,%f,%f];AntXaxis = [%f, %f, %f];AntYaxis=[%f,%f,%f];AntZaxis=[%f,%f,%f];phiAnt=%f;thetaAnt=%f;\n", Xst, Yst, Zst
|
||||
// , AntXaxisX, AntXaxisY, AntXaxisZ
|
||||
// , AntYaxisX, AntYaxisY, AntYaxisZ
|
||||
// , AntZaxisX, AntZaxisY, AntZaxisZ
|
||||
// , phiAnt[idx]
|
||||
// , thetaAnt[idx]
|
||||
//);
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_BillerInterpAntPattern(float* antpattern,
|
||||
float starttheta, float startphi, float dtheta, float dphi,
|
||||
long thetapoints, long phipoints,
|
||||
float* searththeta, float* searchphi, float* searchantpattern,
|
||||
long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
float stheta = searththeta[idx];
|
||||
float sphi = searchphi[idx];
|
||||
float pthetaid = (stheta - starttheta) / dtheta;//
|
||||
float pphiid = (sphi - startphi) / dphi;
|
||||
|
||||
long lasttheta = floorf(pthetaid);
|
||||
long nextTheta = lasttheta + 1;
|
||||
long lastphi = floorf(pphiid);
|
||||
long nextPhi = lastphi + 1;
|
||||
|
||||
if (lasttheta < 0 || nextTheta < 0 || lastphi < 0 || nextPhi < 0 ||
|
||||
lasttheta >= thetapoints || nextTheta >= thetapoints || lastphi >= phipoints || nextPhi >= phipoints)
|
||||
{
|
||||
searchantpattern[idx] = 0;
|
||||
}
|
||||
else {
|
||||
float x = stheta;
|
||||
float y = sphi;
|
||||
|
||||
float x1 = lasttheta * dtheta + starttheta;
|
||||
float x2 = nextTheta * dtheta + starttheta;
|
||||
float y1 = lastphi * dphi + startphi;
|
||||
float y2 = nextPhi * dphi + startphi;
|
||||
|
||||
float z11 = antpattern[lasttheta * phipoints + lastphi];
|
||||
float z12 = antpattern[lasttheta * phipoints + nextPhi];
|
||||
float z21 = antpattern[nextTheta * phipoints + lastphi];
|
||||
float z22 = antpattern[nextTheta * phipoints + nextPhi];
|
||||
|
||||
|
||||
z11 = powf(10, z11 / 10);
|
||||
z12 = powf(10, z12 / 10);
|
||||
z21 = powf(10, z21 / 10);
|
||||
z22 = powf(10, z22 / 10);
|
||||
|
||||
float GainValue = (z11 * (x2 - x) * (y2 - y)
|
||||
+ z21 * (x - x1) * (y2 - y)
|
||||
+ z12 * (x2 - x) * (y - y1)
|
||||
+ z22 * (x - x1) * (y - y1));
|
||||
GainValue = GainValue / ((x2 - x1) * (y2 - y1));
|
||||
searchantpattern[idx] = GainValue;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
__global__ void CUDA_Test_HelloWorld(float a, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
printf("\nidx:\t %d %d \n", idx, len);
|
||||
}
|
||||
|
||||
|
||||
__global__ void CUDA_RTPC(
|
||||
float antPx, float antPy, float antPz,// 天线坐标
|
||||
float antXaxisX, float antXaxisY, float antXaxisZ,
|
||||
float antYaxisX, float antYaxisY, float antYaxisZ,
|
||||
float antZaxisX, float antZaxisY, float antZaxisZ,
|
||||
float antDirectX, float antDirectY, float antDirectZ,
|
||||
float* demx, float* demy, float* demz, long* demcls,
|
||||
float* demslopex, float* demslopey, float* demslopez, float* demslopeangle,
|
||||
float* Tantpattern, float Tstarttheta, float Tstartphi, float Tdtheta, float Tdphi, long Tthetapoints, long Tphipoints,
|
||||
float* Rantpattern, float Rstarttheta, float Rstartphi, float Rdtheta, float Rdphi, long Rthetapoints, long Rphipoints,
|
||||
float lamda, float fs, float nearrange, float Pt, long Freqnumbers, // 参数
|
||||
CUDASigmaParam* sigma0Paramslist, long sigmaparamslistlen,// 地表覆盖类型-sigma插值对应函数-ulaby
|
||||
cuComplex* outecho, int* d_echoAmpFID,
|
||||
int linecount,int plusepoint) {
|
||||
int idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
//printf("\nidx:\t %d %d %d\n", idx, linecount, plusepoint);
|
||||
if (idx < linecount* plusepoint) {
|
||||
long clsid = demcls[idx];
|
||||
CUDAVector Rs{ antPx,antPy,antPz };
|
||||
CUDAVector Rt{ demx[idx],demy[idx],demz[idx] };
|
||||
CUDAVector Rst{ Rs.x - Rt.x,Rs.y - Rt.y,Rs.z - Rt.z };
|
||||
CUDAVector Vslope{ demslopex[idx],demslopey[idx],demslopez[idx] };
|
||||
float R = GPU_VectorNorm2(Rst); // 斜距
|
||||
float slopeangle = demslopeangle[idx];
|
||||
CUDAVectorEllipsoidal Rtanttheta = GPU_SatelliteAntDirectNormal( // 地面目标在天线的位置
|
||||
Rst.x, Rst.y, Rst.z,
|
||||
antXaxisX, antXaxisY, antXaxisZ,
|
||||
antYaxisX, antYaxisY, antYaxisZ,
|
||||
antZaxisX, antZaxisY, antZaxisZ,
|
||||
antDirectX, antDirectY, antDirectZ);
|
||||
|
||||
float localangle = GPU_CosAngle_VectorA_VectorB(Rst, Vslope); // 距地入射角
|
||||
float sigma = GPU_getSigma0dB(sigma0Paramslist[clsid], localangle * r2d);
|
||||
sigma = powf(10.0, sigma / 10.0);// 后向散射系数
|
||||
//printf("\ntheta: %f\t,%f ,%f ,%f ,%f ,%f ,%f \n", localangle * r2d, sigma0Paramslist[clsid].p1, sigma0Paramslist[clsid].p2, sigma0Paramslist[clsid].p3,
|
||||
// sigma0Paramslist[clsid].p4, sigma0Paramslist[clsid].p5, sigma0Paramslist[clsid].p6);
|
||||
// 发射方向图
|
||||
float transPattern = GPU_BillerInterpAntPattern(Tantpattern,
|
||||
Tstarttheta, Tstartphi, Tdtheta, Tdphi, Tthetapoints, Tphipoints,
|
||||
Rtanttheta.theta, Rtanttheta.phi) * r2d;
|
||||
|
||||
// 接收方向图
|
||||
float receivePattern = GPU_BillerInterpAntPattern(Rantpattern,
|
||||
Rstarttheta, Rstartphi, Rdtheta, Rdphi, Rthetapoints, Rphipoints,
|
||||
Rtanttheta.theta, Rtanttheta.phi) * r2d;
|
||||
// 计算振幅、相位
|
||||
float amp = Pt * transPattern * receivePattern * sigma * (1 / cos(slopeangle) * sin(localangle));
|
||||
amp = amp / (powf(4 * LAMP_CUDA_PI, 2) * powf(R, 4));
|
||||
float phi = (-4 * LAMP_CUDA_PI / lamda) * R;
|
||||
|
||||
// 构建回波
|
||||
cuComplex echophi = make_cuComplex(0, phi);
|
||||
cuComplex echophiexp = cuCexpf(echophi);
|
||||
float timeR = 2 * (R - nearrange) / LIGHTSPEED * fs;
|
||||
long timeID = floorf(timeR);
|
||||
if (timeID < 0 || timeID >= Freqnumbers) {
|
||||
timeID = 0;
|
||||
amp = 0;
|
||||
}
|
||||
else {}
|
||||
|
||||
cuComplex echo;
|
||||
echo.x = echophiexp.x * amp;
|
||||
echo.y = echophiexp.y * amp;
|
||||
outecho[idx] = echo;
|
||||
d_echoAmpFID[idx] = timeID;
|
||||
}
|
||||
|
||||
|
||||
}
|
||||
|
||||
|
||||
__global__ void CUDA_TBPImage(
|
||||
float* antPx, float* antPy, float* antPz,
|
||||
float* imgx, float* imgy, float* imgz,
|
||||
cuComplex* echoArr, cuComplex* imgArr,
|
||||
float freq, float fs, float Rnear, float Rfar,
|
||||
long rowcount, long colcount,
|
||||
long prfid, long freqcount
|
||||
) {
|
||||
int idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
//printf("\nidx:\t %d %d %d\n", idx, linecount, plusepoint);
|
||||
if (idx < rowcount * colcount) {
|
||||
float R = sqrtf(powf(antPx[prfid] - imgx[idx], 2) + powf(antPy[prfid] - imgy[idx], 2) + powf(antPz[prfid] - imgz[idx], 2));
|
||||
float Ridf = ((R - Rnear) * 2 / LIGHTSPEED) * fs;
|
||||
long Rid = floorf(Ridf);
|
||||
if(Rid <0|| Rid >= freqcount){}
|
||||
else {
|
||||
float factorj = freq * 4 * PI / LIGHTSPEED;
|
||||
cuComplex Rphi =cuCexpf(make_cuComplex(0, factorj * R));// 校正项
|
||||
imgArr[idx] = cuCaddf(imgArr[idx], cuCmulf(echoArr[Rid] , Rphi));// 矫正
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
__global__ void CUDA_calculationEcho(float* sigma0, float* TransAnt, float* ReciveAnt,
|
||||
float* localangle, float* R, float* slopeangle,
|
||||
float nearRange, float Fs, float Pt, float lamda, long FreqIDmax,
|
||||
cuComplex* echoArr, long* FreqID,
|
||||
long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
float r = R[idx];
|
||||
float amp = Pt * TransAnt[idx] * ReciveAnt[idx];
|
||||
amp = amp * sigma0[idx];
|
||||
amp = amp / (powf(4 * LAMP_CUDA_PI, 2) * powf(r, 4)); // 反射强度
|
||||
|
||||
// 处理相位
|
||||
float phi = (-4 * LAMP_CUDA_PI / lamda) * r;
|
||||
cuComplex echophi = make_cuComplex(0, phi);
|
||||
cuComplex echophiexp = cuCexpf(echophi);
|
||||
|
||||
float timeR = 2 * (r - nearRange) / LIGHTSPEED * Fs;
|
||||
long timeID = floorf(timeR);
|
||||
if (timeID < 0 || timeID >= FreqIDmax) {
|
||||
timeID = 0;
|
||||
amp = 0;
|
||||
}
|
||||
|
||||
cuComplex echo;
|
||||
echo.x = echophiexp.x * amp;
|
||||
echo.y = echophiexp.y * amp;
|
||||
echoArr[idx] = echo;
|
||||
FreqID[idx] = timeID;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
__global__ void CUDA_AntPatternInterpGain(float* anttheta, float* antphi, float* gain,
|
||||
float* antpattern, float starttheta, float startphi, float dtheta, float dphi, int thetapoints, int phipoints, long len) {
|
||||
int idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
|
||||
if (idx < len) {
|
||||
|
||||
float temptheta = anttheta[idx];
|
||||
float tempphi = antphi[idx];
|
||||
|
||||
|
||||
float antPatternGain = GPU_BillerInterpAntPattern(antpattern,
|
||||
starttheta, startphi, dtheta, dphi, thetapoints, phipoints,
|
||||
temptheta, tempphi) ;
|
||||
gain[idx] = antPatternGain;
|
||||
}
|
||||
}
|
||||
|
||||
//__global__ void Sigma0InterpPixel(long* demcls, float* demslopeangle, CUDASigmaParam* sigma0Paramslist, float* localangle, float* sigma0list, long sigmaparamslistlen, long len)
|
||||
//{
|
||||
// long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
// if (idx < len) {
|
||||
// long clsid = demcls[idx];
|
||||
// if(clsid<=)
|
||||
// sigma0list[idx] = 0;
|
||||
// }
|
||||
//}
|
||||
|
||||
|
||||
__global__ void CUDA_InterpSigma(
|
||||
long* demcls, float* sigmaAmp, float* localanglearr, long len,
|
||||
CUDASigmaParam* sigma0Paramslist, long sigmaparamslistlen) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
long clsid = demcls[idx];
|
||||
float localangle = localanglearr[idx] * r2d;
|
||||
CUDASigmaParam tempsigma = sigma0Paramslist[clsid];
|
||||
//printf("cls:%d;localangle=%f;\n",clsid, localangle);
|
||||
|
||||
if (localangle < 0 || localangle >= 90) {
|
||||
sigmaAmp[idx] = 0;
|
||||
}
|
||||
else {}
|
||||
|
||||
if (abs(tempsigma.p1)< PRECISIONTOLERANCE&&
|
||||
abs(tempsigma.p2) < PRECISIONTOLERANCE &&
|
||||
abs(tempsigma.p3) < PRECISIONTOLERANCE &&
|
||||
abs(tempsigma.p4) < PRECISIONTOLERANCE&&
|
||||
abs(tempsigma.p5) < PRECISIONTOLERANCE&&
|
||||
abs(tempsigma.p6) < PRECISIONTOLERANCE
|
||||
) {
|
||||
sigmaAmp[idx] = 0;
|
||||
}
|
||||
else {
|
||||
float sigma = GPU_getSigma0dB(tempsigma, localangle);
|
||||
sigma = powf(10.0, sigma / 10.0);// 后向散射系数
|
||||
//printf("cls:%d;localangle=%f;sigma0=%f;\n", clsid, localangle, sigma);
|
||||
sigmaAmp[idx] = sigma;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
|
||||
|
||||
//错误提示
|
||||
void checkCudaError(cudaError_t err, const char* msg) {
|
||||
if (err != cudaSuccess) {
|
||||
std::cerr << "CUDA error: " << msg << " (" << cudaGetErrorString(err) << ")" << std::endl;
|
||||
exit(EXIT_FAILURE);
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
// 主机参数内存声明
|
||||
extern "C" void* mallocCUDAHost(long memsize) {
|
||||
void* ptr;
|
||||
cudaMallocHost(&ptr, memsize);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("mallocCUDAHost CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
return ptr;
|
||||
}
|
||||
|
||||
// 主机参数内存释放
|
||||
extern "C" void FreeCUDAHost(void* ptr) {
|
||||
cudaFreeHost(ptr);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("FreeCUDAHost CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
// GPU参数内存声明
|
||||
extern "C" void* mallocCUDADevice(long memsize) {
|
||||
void* ptr;
|
||||
cudaMalloc(&ptr, memsize);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("mallocCUDADevice CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
return ptr;
|
||||
}
|
||||
|
||||
// GPU参数内存释放
|
||||
extern "C" void FreeCUDADevice(void* ptr) {
|
||||
cudaFree(ptr);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("FreeCUDADevice CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
// GPU 内存数据转移
|
||||
extern "C" void HostToDevice(void* hostptr, void* deviceptr, long memsize) {
|
||||
cudaMemcpy(deviceptr, hostptr, memsize, cudaMemcpyHostToDevice);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("HostToDevice CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void DeviceToHost(void* hostptr, void* deviceptr, long memsize) {
|
||||
cudaMemcpy(hostptr, deviceptr, memsize, cudaMemcpyDeviceToHost);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("DeviceToHost CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void CUDATestHelloWorld(float a,long len) {
|
||||
// 设置 CUDA 核函数的网格和块的尺寸
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_Test_HelloWorld << <numBlocks, blockSize >> > (a, len);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("FreeCUDADevice CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
void CUDATBPImage(float* antPx, float* antPy, float* antPz,
|
||||
float* imgx, float* imgy, float* imgz,
|
||||
cuComplex* echoArr, cuComplex* imgArr,
|
||||
float freq, float fs, float Rnear, float Rfar,
|
||||
long rowcount, long colcount,
|
||||
long prfid, long freqcount)
|
||||
{
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (rowcount * colcount + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
//printf("\nCUDA_RTPC_SiglePRF blockSize:%d ,numBlock:%d\n",blockSize,numBlocks);
|
||||
// 调用 CUDA 核函数 CUDA_RTPC_Kernel
|
||||
|
||||
CUDA_TBPImage << <numBlocks, blockSize >> > (
|
||||
antPx, antPy, antPz,
|
||||
imgx, imgy, imgz,
|
||||
echoArr, imgArr,
|
||||
freq, fs, Rnear, Rfar,
|
||||
rowcount, colcount,
|
||||
prfid, freqcount
|
||||
);
|
||||
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDATBPImage CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void distanceAB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* R, long len) {
|
||||
// 设置 CUDA 核函数的网格和块的尺寸
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_DistanceAB << <numBlocks, blockSize >> > (Ax, Ay, Az, Bx, By, Bz, R, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void BdistanceAs(float* Ax, float* Ay, float* Az, float Bx, float By, float Bz, float* R, long len) {
|
||||
// 设置 CUDA 核函数的网格和块的尺寸
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_B_DistanceA << <numBlocks, blockSize >> > (Ax, Ay, Az, Bx, By, Bz, R, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void make_VectorA_B(float sX, float sY, float sZ, float* tX, float* tY, float* tZ, float* RstX, float* RstY, float* RstZ, long len) {
|
||||
// 设置 CUDA 核函数的网格和块的尺寸
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_make_VectorA_B << <numBlocks, blockSize >> > (sX, sY, sZ, tX, tY, tZ, RstX, RstY, RstZ, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void Norm_Vector(float* Vx, float* Vy, float* Vz, float* R, long len) {
|
||||
// 设置 CUDA 核函数的网格和块的尺寸
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_Norm_Vector << <numBlocks, blockSize >> > (Vx, Vy, Vz, R, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void cosAngle_VA_AB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* anglecos, long len) {
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_cosAngle_VA_AB << <numBlocks, blockSize >> > (Ax, Ay, Az, Bx, By, Bz, anglecos, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void SatelliteAntDirectNormal(float* RstX, float* RstY, float* RstZ,
|
||||
float antXaxisX, float antXaxisY, float antXaxisZ,
|
||||
float antYaxisX, float antYaxisY, float antYaxisZ,
|
||||
float antZaxisX, float antZaxisY, float antZaxisZ,
|
||||
float antDirectX, float antDirectY, float antDirectZ,
|
||||
float* thetaAnt, float* phiAnt
|
||||
, long len) {
|
||||
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_SatelliteAntDirectNormal << <numBlocks, blockSize >> > (RstX, RstY, RstZ,
|
||||
antXaxisX, antXaxisY, antXaxisZ,
|
||||
antYaxisX, antYaxisY, antYaxisZ,
|
||||
antZaxisX, antZaxisY, antZaxisZ,
|
||||
antDirectX, antDirectY, antDirectZ,
|
||||
thetaAnt, phiAnt
|
||||
, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
|
||||
}
|
||||
|
||||
extern "C" void AntPatternInterpGain(float* anttheta, float* antphi, float* gain,
|
||||
float* antpattern, float starttheta, float startphi, float dtheta, float dphi, int thetapoints, int phipoints, long len) {
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
//printf("\nCUDA_RTPC_SiglePRF blockSize:%d ,numBlock:%d\n", blockSize, numBlocks);
|
||||
|
||||
CUDA_AntPatternInterpGain << <numBlocks, blockSize >> > ( anttheta,antphi, gain,
|
||||
antpattern,
|
||||
starttheta, startphi, dtheta, dphi, thetapoints, phipoints,
|
||||
len);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
|
||||
extern "C" void CUDARTPCPRF(float antPx, long len) {
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
printf("\nCUDA_RTPC_SiglePRF blockSize:%d ,numBlock:%d\n", blockSize, numBlocks);
|
||||
CUDA_Test_HelloWorld << <numBlocks, blockSize >> > (antPx, len);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
|
||||
extern "C" void calculationEcho(float* sigma0, float* TransAnt, float* ReciveAnt,
|
||||
float* localangle, float* R, float* slopeangle,
|
||||
float nearRange, float Fs, float pt, float lamda, long FreqIDmax,
|
||||
cuComplex* echoAmp, long* FreqID,
|
||||
long len)
|
||||
{
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_calculationEcho << <numBlocks, blockSize >> > (sigma0, TransAnt, ReciveAnt,
|
||||
localangle, R, slopeangle,
|
||||
nearRange, Fs, pt, lamda, FreqIDmax,
|
||||
echoAmp, FreqID,
|
||||
len);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
extern "C" void CUDA_RTPC_SiglePRF(
|
||||
float antPx, float antPy, float antPZ,
|
||||
float antXaxisX, float antXaxisY, float antXaxisZ,
|
||||
float antYaxisX, float antYaxisY, float antYaxisZ,
|
||||
float antZaxisX, float antZaxisY, float antZaxisZ,
|
||||
float antDirectX, float antDirectY, float antDirectZ,
|
||||
float* demx, float* demy, float* demz, long* demcls,
|
||||
float* demslopex, float* demslopey, float* demslopez, float* demslopeangle,
|
||||
float* Tantpattern, float Tstarttheta, float Tstartphi, float Tdtheta, float Tdphi, int Tthetapoints, int Tphipoints,
|
||||
float* Rantpattern, float Rstarttheta, float Rstartphi, float Rdtheta, float Rdphi, int Rthetapoints, int Rphipoints,
|
||||
float lamda, float fs, float nearrange, float Pt, int Freqnumbers,
|
||||
CUDASigmaParam* sigma0Paramslist, int sigmaparamslistlen,
|
||||
cuComplex* outecho, int* d_echoAmpFID,
|
||||
int linecount,int colcount) {
|
||||
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (linecount* colcount + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
//printf("\nCUDA_RTPC_SiglePRF blockSize:%d ,numBlock:%d\n",blockSize,numBlocks);
|
||||
// 调用 CUDA 核函数 CUDA_RTPC_Kernel
|
||||
|
||||
CUDA_RTPC << <numBlocks, blockSize >> > (
|
||||
antPx, antPy, antPZ,// 天线坐标
|
||||
antXaxisX, antXaxisY, antXaxisZ, // 天线坐标系
|
||||
antYaxisX, antYaxisY, antYaxisZ, //
|
||||
antZaxisX, antZaxisY, antZaxisZ,
|
||||
antDirectX, antDirectY, antDirectZ,// 天线指向
|
||||
demx, demy, demz,
|
||||
demcls, // 地面坐标
|
||||
demslopex, demslopey, demslopez, demslopeangle,// 地面坡度
|
||||
Tantpattern, Tstarttheta, Tstartphi, Tdtheta, Tdphi, Tthetapoints, Tphipoints,// 天线方向图相关
|
||||
Rantpattern, Rstarttheta, Rstartphi, Rdtheta, Rdphi, Rthetapoints, Rphipoints,// 天线方向图相关
|
||||
lamda, fs, nearrange, Pt, Freqnumbers, // 参数
|
||||
sigma0Paramslist, sigmaparamslistlen,// 地表覆盖类型-sigma插值对应函数-ulaby
|
||||
outecho, d_echoAmpFID,
|
||||
linecount, colcount
|
||||
);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
|
||||
extern "C" void CUDAInterpSigma(
|
||||
long* demcls,float* sigmaAmp, float* localanglearr,long len,
|
||||
CUDASigmaParam* sigma0Paramslist, long sigmaparamslistlen) {// 地表覆盖类型-sigma插值对应函数-ulaby
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_InterpSigma << <numBlocks, blockSize >> > (
|
||||
demcls, sigmaAmp, localanglearr, len,
|
||||
sigma0Paramslist, sigmaparamslistlen
|
||||
);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
|
||||
#endif
|
||||
|
||||
|
||||
|
|
@ -1,452 +0,0 @@
|
|||
#ifndef GPUTOOL_H
|
||||
#define GPUTOOL_H
|
||||
#ifdef __CUDANVCC___
|
||||
#include "BaseConstVariable.h"
|
||||
#include <cuda_runtime.h>
|
||||
#include <device_launch_parameters.h>
|
||||
#include <cublas_v2.h>
|
||||
#include <cuComplex.h>
|
||||
|
||||
#define __CUDADEBUG__
|
||||
|
||||
// 默认显存分布
|
||||
|
||||
|
||||
enum LAMPGPUDATETYPE {
|
||||
LAMP_LONG,
|
||||
LAMP_FLOAT,
|
||||
LAMP_COMPLEXFLOAT
|
||||
};
|
||||
|
||||
|
||||
extern "C" struct CUDASigmaParam {
|
||||
float p1;
|
||||
float p2;
|
||||
float p3;
|
||||
float p4;
|
||||
float p5;
|
||||
float p6;
|
||||
};
|
||||
|
||||
extern "C" struct CUDAVector {
|
||||
float x;
|
||||
float y;
|
||||
float z;
|
||||
};
|
||||
|
||||
extern "C" struct CUDAVectorEllipsoidal {
|
||||
float theta;
|
||||
float phi;
|
||||
float pho;
|
||||
};
|
||||
|
||||
|
||||
|
||||
// GPU 内存函数
|
||||
extern "C" void* mallocCUDAHost( long memsize); // 主机内存声明
|
||||
extern "C" void FreeCUDAHost(void* ptr);
|
||||
extern "C" void* mallocCUDADevice( long memsize); // GPU内存声明
|
||||
extern "C" void FreeCUDADevice(void* ptr);
|
||||
extern "C" void HostToDevice(void* hostptr, void* deviceptr, long memsize);//GPU 内存数据转移 设备 -> GPU
|
||||
extern "C" void DeviceToHost(void* hostptr, void* deviceptr, long memsize);//GPU 内存数据转移 GPU -> 设备
|
||||
|
||||
|
||||
// 仿真所需的常用函数
|
||||
extern "C" void distanceAB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* R, long member);
|
||||
extern "C" void BdistanceAs(float* Ax, float* Ay, float* Az, float Bx, float By, float Bz, float* R, long member);
|
||||
extern "C" void make_VectorA_B(float sX, float sY, float sZ, float* tX, float* tY, float* tZ, float* RstX, float* RstY, float* RstZ, long member);
|
||||
extern "C" void Norm_Vector(float* Vx, float* Vy, float* Vz, float* R, long member);
|
||||
extern "C" void cosAngle_VA_AB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* anglecos, long len);
|
||||
extern "C" void SatelliteAntDirectNormal(float* RstX, float* RstY, float* RstZ, float antXaxisX, float antXaxisY, float antXaxisZ, float antYaxisX, float antYaxisY, float antYaxisZ, float antZaxisX, float antZaxisY, float antZaxisZ, float antDirectX, float antDirectY, float antDirectZ, float* thetaAnt, float* phiAnt, long len);
|
||||
|
||||
extern "C" void AntPatternInterpGain(float* anttheta, float* antphi, float* gain, float* antpattern, float starttheta, float startphi, float dtheta, float dphi, int thetapoints, int phipoints,long len);
|
||||
|
||||
extern "C" void CUDA_RTPC_SiglePRF(
|
||||
float antPx, float antPy, float antPZ,
|
||||
float antXaxisX, float antXaxisY, float antXaxisZ,
|
||||
float antYaxisX, float antYaxisY, float antYaxisZ,
|
||||
float antZaxisX, float antZaxisY, float antZaxisZ,
|
||||
float antDirectX, float antDirectY, float antDirectZ,
|
||||
float* demx, float* demy, float* demz, long* demcls,
|
||||
float* demslopex, float* demslopey, float* demslopez, float* demslopeangle,
|
||||
float* Tantpattern, float Tstarttheta, float Tstartphi, float Tdtheta, float Tdphi, int Tthetapoints, int Tphipoints,
|
||||
float* Rantpattern, float Rstarttheta, float Rstartphi, float Rdtheta, float Rdphi, int Rthetapoints, int Rphipoints,
|
||||
float lamda, float fs, float nearrange, float Pt, int Freqnumbers,
|
||||
CUDASigmaParam* sigma0Paramslist, int sigmaparamslistlen,
|
||||
cuComplex* outecho, int* d_echoAmpFID,
|
||||
int linecount, int colcount
|
||||
);
|
||||
|
||||
extern "C" void CUDARTPCPRF(float antPx, long len);
|
||||
extern "C" void CUDATestHelloWorld(float a, long len);
|
||||
|
||||
|
||||
extern "C" void CUDATBPImage(
|
||||
float* antPx,
|
||||
float* antPy,
|
||||
float* antPz,
|
||||
float* imgx,
|
||||
float* imgy,
|
||||
float* imgz,
|
||||
cuComplex* echoArr,
|
||||
cuComplex* imgArr,
|
||||
float freq, float fs, float Rnear, float Rfar,
|
||||
long rowcount, long colcount,
|
||||
long prfid, long freqcount
|
||||
);
|
||||
|
||||
|
||||
extern "C" void calculationEcho(float* sigma0, float* TransAnt, float* ReciveAnt,
|
||||
float* localangle, float* R, float* slopeangle,
|
||||
float nearRange, float Fs, float pt, float lamda, long FreqIDmax,
|
||||
cuComplex* echoAmp, long* FreqID,
|
||||
long len);
|
||||
|
||||
|
||||
extern "C" void CUDAInterpSigma(
|
||||
long* demcls, float* sigmaAmp, float* localanglearr, long len,
|
||||
CUDASigmaParam* sigma0Paramslist, long sigmaparamslistlen);
|
||||
|
||||
#endif
|
||||
#endif
|
||||
|
||||
|
||||
|
||||
|
||||
/**
|
||||
*
|
||||
|
||||
|
||||
|
||||
double* databuffer = new double[nXSize * nYSize * 2];
|
||||
poBand->RasterIO(GF_Read, start_col, start_row, cols_count, rows_count, databuffer, cols_count,
|
||||
rows_count, GDT_CFloat64, 0, 0);
|
||||
GDALClose((GDALDatasetH)poDataset);
|
||||
Eigen::MatrixXcd rasterData(nYSize, nXSize); // 使用Eigen的MatrixXcd
|
||||
for(size_t i = 0; i < nYSize; i++) {
|
||||
for(size_t j = 0; j < nXSize; j++) {
|
||||
rasterData(i, j) = std::complex<double>(databuffer[i * nXSize * 2 + j * 2],
|
||||
databuffer[i * nXSize * 2 + j * 2 + 1]);
|
||||
}
|
||||
}
|
||||
|
||||
delete[] databuffer;
|
||||
|
||||
|
||||
|
||||
gdalImage demxyz(demxyzPath);// 地面点坐标
|
||||
gdalImage demlandcls(this->LandCoverPath);// 地表覆盖类型
|
||||
gdalImage demsloperxyz(this->demsloperPath);// 地面坡向
|
||||
|
||||
|
||||
omp_lock_t lock; // 定义锁
|
||||
omp_init_lock(&lock); // 初始化锁
|
||||
long start_ids = 1250;
|
||||
for (start_ids = 1; start_ids < demxyz.height; start_ids = start_ids + line_invert) { // 8+ 17 + 0.3 MB
|
||||
QDateTime current = QDateTime::currentDateTime();
|
||||
long pluseStep = Memory1MB * 100 / 3 / PlusePoint;
|
||||
if (pluseStep * num_thread * 3 > this->PluseCount) {
|
||||
pluseStep = this->PluseCount / num_thread / 3;
|
||||
}
|
||||
pluseStep = pluseStep > 50 ? pluseStep : 50;
|
||||
|
||||
|
||||
qDebug() << current.toString("yyyy-MM-dd HH:mm:ss.zzz") << " \tstart \t " << start_ids << " - " << start_ids + line_invert << "\t" << demxyz.height << "\t pluseCount:\t" << pluseStep;
|
||||
// 文件读取
|
||||
Eigen::MatrixXd dem_x = demxyz.getData(start_ids - 1, 0, line_invert + 1, demxyz.width, 1); //
|
||||
Eigen::MatrixXd dem_y = demxyz.getData(start_ids - 1, 0, line_invert + 1, demxyz.width, 2); //
|
||||
Eigen::MatrixXd dem_z = demxyz.getData(start_ids - 1, 0, line_invert + 1, demxyz.width, 3); //
|
||||
|
||||
// 地表覆盖
|
||||
std::shared_ptr<long> dem_landcls = readDataArr<long>(demlandcls, start_ids - 1, 0, line_invert + 1, demxyz.width, 1, GDALREADARRCOPYMETHOD::VARIABLEMETHOD); // 地表覆盖类型
|
||||
long* dem_landcls_ptr = dem_landcls.get();
|
||||
double localAngle = 30.0;
|
||||
|
||||
bool sigmaNoZeroFlag = true;
|
||||
for (long ii = 0; ii < dem_x.rows(); ii++) {
|
||||
for (long jj = 0; jj < dem_y.cols(); jj++) {
|
||||
if (0 != this->SigmaDatabasePtr->getAmp(dem_landcls_ptr[dem_x.cols() * ii + jj], localAngle, polartype)) {
|
||||
sigmaNoZeroFlag = false;
|
||||
break;
|
||||
}
|
||||
}
|
||||
if (!sigmaNoZeroFlag) {
|
||||
break;
|
||||
}
|
||||
}
|
||||
if (sigmaNoZeroFlag) {
|
||||
continue;
|
||||
}
|
||||
|
||||
//#ifdef DEBUGSHOWDIALOG
|
||||
// dialog->load_double_MatrixX_data(dem_z, "dem_z");
|
||||
//#endif
|
||||
|
||||
|
||||
Eigen::MatrixXd demsloper_x = demsloperxyz.getData(start_ids - 1, 0, line_invert + 1, demxyz.width, 1); //
|
||||
Eigen::MatrixXd demsloper_y = demsloperxyz.getData(start_ids - 1, 0, line_invert + 1, demxyz.width, 2); //
|
||||
Eigen::MatrixXd demsloper_z = demsloperxyz.getData(start_ids - 1, 0, line_invert + 1, demxyz.width, 3); //
|
||||
Eigen::MatrixXd sloperAngle = demsloperxyz.getData(start_ids - 1, 0, line_invert + 1, demxyz.width, 4); //
|
||||
sloperAngle = sloperAngle.array() * T180_PI;
|
||||
|
||||
long dem_rows = dem_x.rows();
|
||||
long dem_cols = dem_x.cols();
|
||||
|
||||
long freqidx = 0;//
|
||||
|
||||
|
||||
|
||||
#ifdef DEBUGSHOWDIALOG
|
||||
ImageShowDialogClass* dialog = new ImageShowDialogClass(nullptr);
|
||||
dialog->show();
|
||||
|
||||
Eigen::MatrixXd landaArr = Eigen::MatrixXd::Zero(dem_rows, dem_cols);
|
||||
for (long i = 0; i < dem_rows; i++) {
|
||||
for (long j = 0; j < dem_cols; j++) {
|
||||
landaArr(i, j) = dem_landcls.get()[i * dem_cols + j];
|
||||
}
|
||||
}
|
||||
dialog->load_double_MatrixX_data(landaArr, "landCover");
|
||||
#endif
|
||||
//qDebug() << " pluse bolck size :\t " << pluseStep << " all size:\t" << this->PluseCount;
|
||||
long processNumber = 0;
|
||||
|
||||
#pragma omp parallel for
|
||||
for (long startprfidx = 0; startprfidx < this->PluseCount; startprfidx = startprfidx + pluseStep) { // 17 + 0.3 MB
|
||||
long prfcount_step = startprfidx + pluseStep < this->PluseCount ? pluseStep : this->PluseCount - startprfidx;
|
||||
Eigen::MatrixXcd echoPluse = Eigen::MatrixXcd::Zero(prfcount_step, PlusePoint); // 当前脉冲的回波积分情况
|
||||
// 内存预分配
|
||||
Eigen::MatrixXd Rst_x = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd Rst_y = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd Rst_z = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd R = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd localangle = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd Vst_x = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd Vst_y = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd Vst_z = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd fde = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd fr = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd Rx = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd sigam = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd echoAmp = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols()).array() + Pt;
|
||||
Eigen::MatrixXd Rphi = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd TimeRange = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd TransAnt = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd ReciveAnt = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
|
||||
Eigen::MatrixXd AntTheta = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
Eigen::MatrixXd AntPhi = Eigen::MatrixXd::Zero(dem_x.rows(), dem_x.cols());
|
||||
|
||||
double minR = 0, maxR = 0;
|
||||
double minLocalAngle = 0, maxLocalAngle = 0;
|
||||
|
||||
Vector3D Rt = { 0,0,0 };
|
||||
SatelliteOribtNode oRs = SatelliteOribtNode{ 0 };;
|
||||
|
||||
Vector3D p0 = {}, slopeVector = {}, sateAntDirect = {};
|
||||
Vector3D Rs = {}, Vs = {}, Ast = {};
|
||||
SatelliteAntDirect antdirectNode = {};
|
||||
std::complex<double> echofreq;
|
||||
std::complex<double> Imag1(0, 1);
|
||||
double TAntPattern = 1; // 发射天线方向图
|
||||
double RAntPanttern = 1;// 接收天线方向图
|
||||
double maxechoAmp = 1;
|
||||
double tempAmp = 1;
|
||||
for (long prfidx = 0; prfidx < prfcount_step; prfidx++)
|
||||
{
|
||||
oRs = sateOirbtNodes[prfidx + startprfidx];
|
||||
|
||||
// 计算天线方向图
|
||||
for (long jj = 1; jj < dem_cols - 1; jj++) {
|
||||
for (long ii = 1; ii < dem_rows - 1; ii++) {
|
||||
p0.x = dem_x(ii, jj);
|
||||
p0.y = dem_y(ii, jj);
|
||||
p0.z = dem_z(ii, jj);
|
||||
this->TaskSetting->getSatelliteAntDirectNormal(oRs, p0, antdirectNode);
|
||||
//antdirectNode.ThetaAnt = antdirectNode.ThetaAnt * r2d;
|
||||
//antdirectNode.PhiAnt = antdirectNode.PhiAnt * r2d;
|
||||
AntTheta(ii, jj) = antdirectNode.ThetaAnt * r2d;
|
||||
AntPhi(ii, jj) = antdirectNode.PhiAnt * r2d;
|
||||
}
|
||||
}
|
||||
|
||||
// 计算发射天线方向图
|
||||
for (long jj = 1; jj < dem_cols - 1; jj++) {
|
||||
for (long ii = 1; ii < dem_rows - 1; ii++) {
|
||||
TransformPattern->getGainLinear(AntTheta(ii, jj), AntPhi(ii, jj), TransAnt(ii, jj));
|
||||
//TransAnt(ii, jj) = TAntPattern;
|
||||
}
|
||||
}
|
||||
|
||||
// 计算接收天线方向图
|
||||
for (long jj = 1; jj < dem_cols - 1; jj++) {
|
||||
for (long ii = 1; ii < dem_rows - 1; ii++) {
|
||||
TransformPattern->getGainLinear(AntTheta(ii, jj), AntPhi(ii, jj), ReciveAnt(ii, jj));
|
||||
//ReciveAnt(ii, jj) = RAntPanttern;
|
||||
}
|
||||
}
|
||||
|
||||
// 计算经过增益的能量
|
||||
echoAmp = Pt * TransAnt.array() * ReciveAnt.array();
|
||||
|
||||
maxechoAmp = echoAmp.maxCoeff();
|
||||
if (std::abs(maxechoAmp) < PRECISIONTOLERANCE) { // 这种情况下，不在合成孔径范围中
|
||||
continue;
|
||||
}
|
||||
|
||||
Rs.x = sateOirbtNodes[prfidx + startprfidx].Px; // 卫星位置
|
||||
Rs.y = sateOirbtNodes[prfidx + startprfidx].Py;
|
||||
Rs.z = sateOirbtNodes[prfidx + startprfidx].Pz;
|
||||
|
||||
Vs.x = sateOirbtNodes[prfidx + startprfidx].Vx; // 卫星速度
|
||||
Vs.y = sateOirbtNodes[prfidx + startprfidx].Vy;
|
||||
Vs.z = sateOirbtNodes[prfidx + startprfidx].Vz;
|
||||
|
||||
Ast.x = sateOirbtNodes[prfidx + startprfidx].AVx;// 卫星加速度
|
||||
Ast.y = sateOirbtNodes[prfidx + startprfidx].AVy;
|
||||
Ast.z = sateOirbtNodes[prfidx + startprfidx].AVz;
|
||||
|
||||
Rst_x = Rs.x - dem_x.array(); // Rst = Rs - Rt;
|
||||
Rst_y = Rs.y - dem_y.array();
|
||||
Rst_z = Rs.z - dem_z.array();
|
||||
R = (Rst_x.array().pow(2) + Rst_y.array().pow(2) + Rst_z.array().pow(2)).array().sqrt(); // R
|
||||
|
||||
minR = R.minCoeff();
|
||||
maxR = R.maxCoeff();
|
||||
//qDebug() << "minR:\t" << minR << " maxR:\t" << maxR;
|
||||
if (maxR<NearRange || minR>FarRange) {
|
||||
continue;
|
||||
}
|
||||
else {}
|
||||
|
||||
// getCosAngle
|
||||
// double c = dot(a, b) / (getlength(a) * getlength(b));
|
||||
// return acos(c > 1 ? 1 : c < -1 ? -1 : c) * r2d;
|
||||
// localangle = getCosAngle(Rst, slopeVector) * T180_PI; // 注意这个只能实时计算，因为非实时计算代价太大
|
||||
localangle = (Rst_x.array() * demsloper_x.array() + Rst_y.array() * demsloper_y.array() + Rst_z.array() * demsloper_z.array()).array(); // dot(a, b)
|
||||
localangle = localangle.array() / R.array();
|
||||
localangle = localangle.array() / (demsloper_x.array().pow(2) + demsloper_y.array().pow(2) + demsloper_z.array().pow(2)).array().sqrt().array();
|
||||
localangle = localangle.array().acos(); // 弧度值
|
||||
|
||||
minLocalAngle = localangle.minCoeff();
|
||||
maxLocalAngle = localangle.maxCoeff();
|
||||
if (maxLocalAngle<0 || minLocalAngle>PI / 2) {
|
||||
continue;
|
||||
}
|
||||
else {}
|
||||
|
||||
//Vst_x = Vs.x + 1 * earthRoute * dem_y.array(); // Vst = Vs - Vt;
|
||||
//Vst_y = Vs.y - 1 * earthRoute * dem_x.array();
|
||||
//Vst_z = Vs.z - Eigen::MatrixXd::Zero(dem_x.rows(), dem_y.cols()).array();
|
||||
|
||||
//// 计算多普勒中心频率 Rst, Vst : ( - 2 / lamda) * dot(Rs - Rt, Vs - Vt) / R; // 星载合成孔径雷达原始回波数据模拟研究 3.18
|
||||
//fde = (-2 / lamda) * (Rst_x.array() * Vst_x.array() + Rst_y.array() * Vst_y.array() + Rst_z.array() * Vst_z.array()).array() / (R.array());
|
||||
//// 计算多普勒频率斜率 // 星载合成孔径雷达原始回波数据模拟研究 3.19
|
||||
//// -(2/lamda)*( dot(Vs - Vt, Vs - Vt)/R + dot(Ast, Rs - Rt)/R - std::pow(dot(Vs - Vt, Rs - Rt),2 )/std::pow(R,3));
|
||||
//fr = (-2 / lamda) *
|
||||
// (Vst_x.array() * Vst_x.array() + Vst_y.array() * Vst_y.array() + Vst_z.array() * Vst_z.array()).array() / (R.array()) +
|
||||
// (-2 / lamda) *
|
||||
// (Ast.x * Rst_x.array() + Ast.y * Rst_y.array() + Ast.z * Rst_z.array()).array() / (R.array()) -
|
||||
// (-2 / lamda) *
|
||||
// (Vst_x.array() * Rst_x.array() + Vst_y.array() * Rst_y.array() + Vst_z.array() * Rst_z.array()).array().pow(2) / (R.array().pow(3));
|
||||
// 计算回波
|
||||
Rx = R;// -(lamda / 2) * (fde * TRx + 0.5 * fr * TRx * TRx); // 斜距历程值
|
||||
|
||||
|
||||
// 逐点计算 this->SigmaDatabasePtr->getAmp(covercls, localangle, polartype); // 后向散射系数 HH
|
||||
|
||||
for (long ii = 0; ii < dem_x.rows(); ii++) {
|
||||
for (long jj = 0; jj < dem_y.cols(); jj++) {
|
||||
sigam(ii, jj) = this->SigmaDatabasePtr->getAmp(dem_landcls_ptr[dem_x.cols() * ii + jj], localangle(ii, jj) * r2d, polartype);
|
||||
}
|
||||
}
|
||||
|
||||
if (sigam.maxCoeff() > 0) {}
|
||||
else {
|
||||
continue;
|
||||
}
|
||||
// projArea = 1 / std::cos(sloperAngle) * std::sin(localangle); // 投影面积系数，单位投影面积 1m x 1m --注意这里是假设，后期再补充
|
||||
// echoAmp = projArea*TAntPattern * RAntPanttern * sigam / (4 * PI * R * R);
|
||||
|
||||
echoAmp = echoAmp.array() * sigam.array() * (1 / sloperAngle.array().cos() * localangle.array().sin()); // 反射强度
|
||||
echoAmp = echoAmp.array() / (4 * PI * R.array().pow(2)); // 距离衰减
|
||||
|
||||
Rphi = -4 * PI / lamda * Rx.array();// 距离徙动相位
|
||||
// 积分
|
||||
TimeRange = ((2 * R.array() / LIGHTSPEED - TimgNearRange).array() * Fs).array();
|
||||
double localAnglepoint = -1;
|
||||
long prf_freq_id = 0;
|
||||
|
||||
|
||||
for (long jj = 1; jj < dem_cols - 1; jj++) {
|
||||
for (long ii = 1; ii < dem_rows - 1; ii++) {
|
||||
prf_freq_id = std::floor(TimeRange(ii, jj));
|
||||
if (prf_freq_id < 0 || prf_freq_id >= PlusePoint || localangle(ii, jj) < 0 || localangle(ii, jj) > PI / 2 || echoAmp(ii, jj) == 0) {
|
||||
continue;
|
||||
}
|
||||
echofreq = echoAmp(ii, jj) * std::exp(Rphi(ii, jj) * Imag1);
|
||||
echoPluse(prfidx, prf_freq_id) = echoPluse(prfidx, prf_freq_id) + echofreq;
|
||||
}
|
||||
}
|
||||
|
||||
#ifdef DEBUGSHOWDIALOG
|
||||
ImageShowDialogClass* localangledialog = new ImageShowDialogClass(dialog);
|
||||
localangledialog->show();
|
||||
localangledialog->load_double_MatrixX_data(localangle.array() * r2d, "localangle");
|
||||
|
||||
|
||||
ImageShowDialogClass* sigamdialog = new ImageShowDialogClass(dialog);
|
||||
sigamdialog->show();
|
||||
sigamdialog->load_double_MatrixX_data(TimeRange, "TimeRange");
|
||||
|
||||
|
||||
ImageShowDialogClass* ampdialog = new ImageShowDialogClass(dialog);
|
||||
ampdialog->show();
|
||||
ampdialog->load_double_MatrixX_data(echoAmp, "echoAmp");
|
||||
|
||||
Eigen::MatrixXd echoPluseamp = echoPluse.array().abs().cast<double>().array();
|
||||
ImageShowDialogClass* echoampdialog = new ImageShowDialogClass(dialog);
|
||||
echoampdialog->show();
|
||||
echoampdialog->load_double_MatrixX_data(echoPluseamp, "echoPluseamp");
|
||||
|
||||
|
||||
dialog->exec();
|
||||
#endif
|
||||
//qDebug() << QDateTime::currentDateTime().toString("yyyy-MM-dd HH:mm:ss.zzz") << " end " << prfidx;
|
||||
}
|
||||
//qDebug() << QDateTime::currentDateTime().toString("yyyy-MM-dd HH:mm:ss.zzz")<<" step "<< prfcount_step;
|
||||
|
||||
omp_set_lock(&lock); // 回波整体赋值处理
|
||||
for (long prfidx = 0; prfidx < prfcount_step; prfidx++) {
|
||||
for (long freqidx = 0; freqidx < PlusePoint; freqidx++)
|
||||
{
|
||||
//qDebug() << prfidx << " " << freqidx << " " << echoPluse(prfidx, freqidx).real() << " + " << echoPluse(prfidx, freqidx).imag() << " j";
|
||||
echo.get()[(prfidx + startprfidx) * PlusePoint + freqidx] = echo.get()[(prfidx + startprfidx) * PlusePoint + freqidx] + echoPluse(prfidx, freqidx);
|
||||
}
|
||||
}
|
||||
//this->EchoSimulationData->saveEchoArr(echo, 0, PluseCount);
|
||||
omp_unset_lock(&lock); // 解锁
|
||||
//qDebug() << QDateTime::currentDateTime().toString("yyyy-MM-dd HH:mm:ss.zzz") << " step 2" << prfcount_step;
|
||||
}
|
||||
|
||||
omp_set_lock(&lock); // 保存文件
|
||||
processNumber = processNumber + pluseStep;
|
||||
|
||||
this->EchoSimulationData->saveEchoArr(echo, 0, PluseCount);
|
||||
omp_unset_lock(&lock); // 解锁
|
||||
|
||||
qDebug() << QDateTime::currentDateTime().toString("yyyy-MM-dd HH:mm:ss.zzz") << " \t " << start_ids << "\t--\t " << start_ids + line_invert << "\t/\t" << demxyz.height;
|
||||
|
||||
}
|
||||
omp_destroy_lock(&lock); // 销毁锁
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
*/
|
||||
|
|
@ -0,0 +1,464 @@
|
|||
|
||||
|
||||
#include <iostream>
|
||||
#include <memory>
|
||||
#include <cmath>
|
||||
#include <complex>
|
||||
#include <device_launch_parameters.h>
|
||||
#include <cuda_runtime.h>
|
||||
#include <cublas_v2.h>
|
||||
#include <cuComplex.h>
|
||||
|
||||
#include "BaseConstVariable.h"
|
||||
#include "GPURTPC.cuh"
|
||||
|
||||
#ifdef __CUDANVCC___
|
||||
|
||||
|
||||
__device__ float GPU_getSigma0dB(CUDASigmaParam param, float theta) {//线性值
|
||||
float sigma= param.p1 + param.p2 * exp(-param.p3 * theta) + param.p4 * cos(param.p5 * theta + param.p6);
|
||||
return sigma;
|
||||
}
|
||||
|
||||
|
||||
__device__ CUDAVectorEllipsoidal GPU_SatelliteAntDirectNormal(float RstX, float RstY, float RstZ,
|
||||
float antXaxisX, float antXaxisY, float antXaxisZ,
|
||||
float antYaxisX, float antYaxisY, float antYaxisZ,
|
||||
float antZaxisX, float antZaxisY, float antZaxisZ,
|
||||
float antDirectX, float antDirectY, float antDirectZ
|
||||
) {
|
||||
CUDAVectorEllipsoidal result{ 0,0,-1 };
|
||||
float Xst = -1 * RstX; // 卫星 --> 地面
|
||||
float Yst = -1 * RstY;
|
||||
float Zst = -1 * RstZ;
|
||||
float AntXaxisX = antXaxisX;
|
||||
float AntXaxisY = antXaxisY;
|
||||
float AntXaxisZ = antXaxisZ;
|
||||
float AntYaxisX = antYaxisX;
|
||||
float AntYaxisY = antYaxisY;
|
||||
float AntYaxisZ = antYaxisZ;
|
||||
float AntZaxisX = antZaxisX;
|
||||
float AntZaxisY = antZaxisY;
|
||||
float AntZaxisZ = antZaxisZ;
|
||||
// 天线指向在天线坐标系下的值
|
||||
float Xant = (Xst * (AntYaxisY * AntZaxisZ - AntYaxisZ * AntZaxisY) + Xst * (AntXaxisZ * AntZaxisY - AntXaxisY * AntZaxisZ) + Xst * (AntXaxisY * AntYaxisZ - AntXaxisZ * AntYaxisY)) / (AntXaxisX * (AntYaxisY * AntZaxisZ - AntZaxisY * AntYaxisZ) - AntYaxisX * (AntXaxisY * AntZaxisZ - AntXaxisZ * AntZaxisY) + AntZaxisX * (AntXaxisY * AntYaxisZ - AntXaxisZ * AntYaxisY));
|
||||
float Yant = (Yst * (AntYaxisZ * AntZaxisX - AntYaxisX * AntZaxisZ) + Yst * (AntXaxisX * AntZaxisZ - AntXaxisZ * AntZaxisX) + Yst * (AntYaxisX * AntXaxisZ - AntXaxisX * AntYaxisZ)) / (AntXaxisX * (AntYaxisY * AntZaxisZ - AntZaxisY * AntYaxisZ) - AntYaxisX * (AntXaxisY * AntZaxisZ - AntXaxisZ * AntZaxisY) + AntZaxisX * (AntXaxisY * AntYaxisZ - AntXaxisZ * AntYaxisY));
|
||||
float Zant = (Zst * (AntYaxisX * AntZaxisY - AntYaxisY * AntZaxisX) + Zst * (AntXaxisY * AntZaxisX - AntXaxisX * AntZaxisY) + Zst * (AntXaxisX * AntYaxisY - AntYaxisX * AntXaxisY)) / (AntXaxisX * (AntYaxisY * AntZaxisZ - AntZaxisY * AntYaxisZ) - AntYaxisX * (AntXaxisY * AntZaxisZ - AntXaxisZ * AntZaxisY) + AntZaxisX * (AntXaxisY * AntYaxisZ - AntXaxisZ * AntYaxisY));
|
||||
// 计算theta 与 phi
|
||||
float Norm = sqrtf(Xant * Xant + Yant * Yant + Zant * Zant); // 计算 pho
|
||||
float ThetaAnt = acosf(Zant / Norm); // theta 与 Z轴的夹角
|
||||
float YsinTheta = Yant / sinf(ThetaAnt);
|
||||
float PhiAnt = (YsinTheta / abs(YsinTheta)) * acosf(Xant / (Norm * sinf(ThetaAnt)));
|
||||
result.theta = ThetaAnt;
|
||||
result.phi = PhiAnt;
|
||||
result.pho = Norm;
|
||||
return result;
|
||||
}
|
||||
|
||||
__device__ float GPU_BillerInterpAntPattern(float* antpattern,
|
||||
float starttheta, float startphi, float dtheta, float dphi,
|
||||
long thetapoints, long phipoints,
|
||||
float searththeta, float searchphi) {
|
||||
float stheta = searththeta;
|
||||
float sphi = searchphi;
|
||||
if (stheta > 90) {
|
||||
return 0;
|
||||
}
|
||||
else {}
|
||||
|
||||
|
||||
float pthetaid = (stheta - starttheta) / dtheta;//
|
||||
float pphiid = (sphi - startphi) / dphi;
|
||||
|
||||
long lasttheta = floorf(pthetaid);
|
||||
long nextTheta = lasttheta + 1;
|
||||
long lastphi = floorf(pphiid);
|
||||
long nextPhi = lastphi + 1;
|
||||
|
||||
|
||||
if (lasttheta < 0 || nextTheta < 0 || lastphi < 0 || nextPhi < 0 ||
|
||||
lasttheta >= thetapoints || nextTheta >= thetapoints || lastphi >= phipoints || nextPhi >= phipoints)
|
||||
{
|
||||
return 0;
|
||||
}
|
||||
else {
|
||||
float x = stheta;
|
||||
float y = sphi;
|
||||
|
||||
float x1 = lasttheta * dtheta + starttheta;
|
||||
float x2 = nextTheta * dtheta + starttheta;
|
||||
float y1 = lastphi * dphi + startphi;
|
||||
float y2 = nextPhi * dphi + startphi;
|
||||
|
||||
float z11 = antpattern[lasttheta * phipoints + lastphi];
|
||||
float z12 = antpattern[lasttheta * phipoints + nextPhi];
|
||||
float z21 = antpattern[nextTheta * phipoints + lastphi];
|
||||
float z22 = antpattern[nextTheta * phipoints + nextPhi];
|
||||
|
||||
|
||||
//z11 = powf(10, z11 / 10); // dB-> 线性
|
||||
//z12 = powf(10, z12 / 10);
|
||||
//z21 = powf(10, z21 / 10);
|
||||
//z22 = powf(10, z22 / 10);
|
||||
|
||||
float GainValue = (z11 * (x2 - x) * (y2 - y)
|
||||
+ z21 * (x - x1) * (y2 - y)
|
||||
+ z12 * (x2 - x) * (y - y1)
|
||||
+ z22 * (x - x1) * (y - y1));
|
||||
GainValue = GainValue / ((x2 - x1) * (y2 - y1));
|
||||
return GainValue;
|
||||
}
|
||||
}
|
||||
|
||||
__device__ cuComplex GPU_calculationEcho(float sigma0, float TransAnt, float ReciveAnt,
|
||||
float localangle, float R, float slopeangle, float Pt, float lamda) {
|
||||
float r = R;
|
||||
float amp = Pt * TransAnt * ReciveAnt;
|
||||
amp = amp * sigma0;
|
||||
amp = amp / (powf(4 * LAMP_CUDA_PI, 2) * powf(r, 4)); // 反射强度
|
||||
float phi = (-4 * LAMP_CUDA_PI / lamda) * r;
|
||||
cuComplex echophi = make_cuComplex(0, phi);
|
||||
cuComplex echophiexp = cuCexpf(echophi);
|
||||
cuComplex echo;
|
||||
echo.x = echophiexp.x * amp;
|
||||
echo.y = echophiexp.y * amp;
|
||||
return echo;
|
||||
}
|
||||
|
||||
|
||||
__global__ void CUDA_SatelliteAntDirectNormal(float* RstX, float* RstY, float* RstZ,
|
||||
float antXaxisX, float antXaxisY, float antXaxisZ,
|
||||
float antYaxisX, float antYaxisY, float antYaxisZ,
|
||||
float antZaxisX, float antZaxisY, float antZaxisZ,
|
||||
float antDirectX, float antDirectY, float antDirectZ,
|
||||
float* thetaAnt, float* phiAnt
|
||||
, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
float Xst = -1 * RstX[idx]; // 卫星 --> 地面
|
||||
float Yst = -1 * RstY[idx];
|
||||
float Zst = -1 * RstZ[idx];
|
||||
float AntXaxisX = antXaxisX;
|
||||
float AntXaxisY = antXaxisY;
|
||||
float AntXaxisZ = antXaxisZ;
|
||||
float AntYaxisX = antYaxisX;
|
||||
float AntYaxisY = antYaxisY;
|
||||
float AntYaxisZ = antYaxisZ;
|
||||
float AntZaxisX = antZaxisX;
|
||||
float AntZaxisY = antZaxisY;
|
||||
float AntZaxisZ = antZaxisZ;
|
||||
|
||||
// 归一化
|
||||
float RstNorm = sqrtf(Xst * Xst + Yst * Yst + Zst * Zst);
|
||||
float AntXaxisNorm = sqrtf(AntXaxisX * AntXaxisX + AntXaxisY * AntXaxisY + AntXaxisZ * AntXaxisZ);
|
||||
float AntYaxisNorm = sqrtf(AntYaxisX * AntYaxisX + AntYaxisY * AntYaxisY + AntYaxisZ * AntYaxisZ);
|
||||
float AntZaxisNorm = sqrtf(AntZaxisX * AntZaxisX + AntZaxisY * AntZaxisY + AntZaxisZ * AntZaxisZ);
|
||||
|
||||
|
||||
float Rx = Xst / RstNorm;
|
||||
float Ry = Yst / RstNorm;
|
||||
float Rz = Zst / RstNorm;
|
||||
float Xx = AntXaxisX / AntXaxisNorm;
|
||||
float Xy = AntXaxisY / AntXaxisNorm;
|
||||
float Xz = AntXaxisZ / AntXaxisNorm;
|
||||
float Yx = AntYaxisX / AntYaxisNorm;
|
||||
float Yy = AntYaxisY / AntYaxisNorm;
|
||||
float Yz = AntYaxisZ / AntYaxisNorm;
|
||||
float Zx = AntZaxisX / AntZaxisNorm;
|
||||
float Zy = AntZaxisY / AntZaxisNorm;
|
||||
float Zz = AntZaxisZ / AntZaxisNorm;
|
||||
|
||||
float Xant = (Rx * Yy * Zz - Rx * Yz * Zy - Ry * Yx * Zz + Ry * Yz * Zx + Rz * Yx * Zy - Rz * Yy * Zx) / (Xx * Yy * Zz - Xx * Yz * Zy - Xy * Yx * Zz + Xy * Yz * Zx + Xz * Yx * Zy - Xz * Yy * Zx);
|
||||
float Yant = -(Rx * Xy * Zz - Rx * Xz * Zy - Ry * Xx * Zz + Ry * Xz * Zx + Rz * Xx * Zy - Rz * Xy * Zx) / (Xx * Yy * Zz - Xx * Yz * Zy - Xy * Yx * Zz + Xy * Yz * Zx + Xz * Yx * Zy - Xz * Yy * Zx);
|
||||
float Zant = (Rx * Xy * Yz - Rx * Xz * Yy - Ry * Xx * Yz + Ry * Xz * Yx + Rz * Xx * Yy - Rz * Xy * Yx) / (Xx * Yy * Zz - Xx * Yz * Zy - Xy * Yx * Zz + Xy * Yz * Zx + Xz * Yx * Zy - Xz * Yy * Zx);
|
||||
|
||||
|
||||
// 计算theta 与 phi
|
||||
float Norm = sqrtf(Xant * Xant + Yant * Yant + Zant * Zant); // 计算 pho
|
||||
float ThetaAnt = acosf(Zant / Norm); // theta 与 Z轴的夹角
|
||||
float PhiAnt = atanf(Yant / Xant); // -pi/2 ~pi/2
|
||||
|
||||
|
||||
if (abs(Yant) < PRECISIONTOLERANCE) { // X轴上
|
||||
PhiAnt = 0;
|
||||
}
|
||||
else if (abs(Xant) < PRECISIONTOLERANCE) { // Y轴上，原点
|
||||
if (Yant > 0) {
|
||||
PhiAnt = PI / 2;
|
||||
}
|
||||
else {
|
||||
PhiAnt = -PI / 2;
|
||||
}
|
||||
}
|
||||
else if (Xant < 0) {
|
||||
if (Yant > 0) {
|
||||
PhiAnt = PI + PhiAnt;
|
||||
}
|
||||
else {
|
||||
PhiAnt = -PI+PhiAnt ;
|
||||
}
|
||||
}
|
||||
else { // Xant>0 X 正轴
|
||||
|
||||
}
|
||||
|
||||
if (isnan(PhiAnt)) {
|
||||
printf("V=[%f,%f,%f];norm=%f;thetaAnt=%f;phiAnt=%f;\n", Xant, Yant, Zant,Norm, ThetaAnt, PhiAnt);
|
||||
}
|
||||
|
||||
//if (abs(ThetaAnt - 0) < PRECISIONTOLERANCE) {
|
||||
// PhiAnt = 0;
|
||||
//}
|
||||
//else {}
|
||||
|
||||
|
||||
thetaAnt[idx] = ThetaAnt*r2d;
|
||||
phiAnt[idx] = PhiAnt*r2d;
|
||||
//printf("Rst=[%f,%f,%f];AntXaxis = [%f, %f, %f];AntYaxis=[%f,%f,%f];AntZaxis=[%f,%f,%f];phiAnt=%f;thetaAnt=%f;\n", Xst, Yst, Zst
|
||||
// , AntXaxisX, AntXaxisY, AntXaxisZ
|
||||
// , AntYaxisX, AntYaxisY, AntYaxisZ
|
||||
// , AntZaxisX, AntZaxisY, AntZaxisZ
|
||||
// , phiAnt[idx]
|
||||
// , thetaAnt[idx]
|
||||
//);
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_BillerInterpAntPattern(float* antpattern,
|
||||
float starttheta, float startphi, float dtheta, float dphi,
|
||||
long thetapoints, long phipoints,
|
||||
float* searththeta, float* searchphi, float* searchantpattern,
|
||||
long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
float stheta = searththeta[idx];
|
||||
float sphi = searchphi[idx];
|
||||
float pthetaid = (stheta - starttheta) / dtheta;//
|
||||
float pphiid = (sphi - startphi) / dphi;
|
||||
|
||||
long lasttheta = floorf(pthetaid);
|
||||
long nextTheta = lasttheta + 1;
|
||||
long lastphi = floorf(pphiid);
|
||||
long nextPhi = lastphi + 1;
|
||||
|
||||
if (lasttheta < 0 || nextTheta < 0 || lastphi < 0 || nextPhi < 0 ||
|
||||
lasttheta >= thetapoints || nextTheta >= thetapoints || lastphi >= phipoints || nextPhi >= phipoints)
|
||||
{
|
||||
searchantpattern[idx] = 0;
|
||||
}
|
||||
else {
|
||||
float x = stheta;
|
||||
float y = sphi;
|
||||
|
||||
float x1 = lasttheta * dtheta + starttheta;
|
||||
float x2 = nextTheta * dtheta + starttheta;
|
||||
float y1 = lastphi * dphi + startphi;
|
||||
float y2 = nextPhi * dphi + startphi;
|
||||
|
||||
float z11 = antpattern[lasttheta * phipoints + lastphi];
|
||||
float z12 = antpattern[lasttheta * phipoints + nextPhi];
|
||||
float z21 = antpattern[nextTheta * phipoints + lastphi];
|
||||
float z22 = antpattern[nextTheta * phipoints + nextPhi];
|
||||
|
||||
|
||||
z11 = powf(10, z11 / 10);
|
||||
z12 = powf(10, z12 / 10);
|
||||
z21 = powf(10, z21 / 10);
|
||||
z22 = powf(10, z22 / 10);
|
||||
|
||||
float GainValue = (z11 * (x2 - x) * (y2 - y)
|
||||
+ z21 * (x - x1) * (y2 - y)
|
||||
+ z12 * (x2 - x) * (y - y1)
|
||||
+ z22 * (x - x1) * (y - y1));
|
||||
GainValue = GainValue / ((x2 - x1) * (y2 - y1));
|
||||
searchantpattern[idx] = GainValue;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_calculationEcho(float* sigma0, float* TransAnt, float* ReciveAnt,
|
||||
float* localangle, float* R, float* slopeangle,
|
||||
float nearRange, float Fs, float Pt, float lamda, long FreqIDmax,
|
||||
cuComplex* echoArr, long* FreqID,
|
||||
long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
float r = R[idx];
|
||||
float amp = Pt * TransAnt[idx] * ReciveAnt[idx];
|
||||
amp = amp * sigma0[idx];
|
||||
amp = amp / (powf(4 * LAMP_CUDA_PI, 2) * powf(r, 4)); // 反射强度
|
||||
|
||||
// 处理相位
|
||||
float phi = (-4 * LAMP_CUDA_PI / lamda) * r;
|
||||
cuComplex echophi = make_cuComplex(0, phi);
|
||||
cuComplex echophiexp = cuCexpf(echophi);
|
||||
|
||||
float timeR = 2 * (r - nearRange) / LIGHTSPEED * Fs;
|
||||
long timeID = floorf(timeR);
|
||||
if (timeID < 0 || timeID >= FreqIDmax) {
|
||||
timeID = 0;
|
||||
amp = 0;
|
||||
}
|
||||
|
||||
cuComplex echo;
|
||||
echo.x = echophiexp.x * amp;
|
||||
echo.y = echophiexp.y * amp;
|
||||
echoArr[idx] = echo;
|
||||
FreqID[idx] = timeID;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
__global__ void CUDA_AntPatternInterpGain(float* anttheta, float* antphi, float* gain,
|
||||
float* antpattern, float starttheta, float startphi, float dtheta, float dphi, int thetapoints, int phipoints, long len) {
|
||||
int idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
|
||||
if (idx < len) {
|
||||
|
||||
float temptheta = anttheta[idx];
|
||||
float tempphi = antphi[idx];
|
||||
|
||||
|
||||
float antPatternGain = GPU_BillerInterpAntPattern(antpattern,
|
||||
starttheta, startphi, dtheta, dphi, thetapoints, phipoints,
|
||||
temptheta, tempphi) ;
|
||||
gain[idx] = antPatternGain;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
__global__ void CUDA_InterpSigma(
|
||||
long* demcls, float* sigmaAmp, float* localanglearr, long len,
|
||||
CUDASigmaParam* sigma0Paramslist, long sigmaparamslistlen) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
long clsid = demcls[idx];
|
||||
float localangle = localanglearr[idx];
|
||||
CUDASigmaParam tempsigma = sigma0Paramslist[clsid];
|
||||
//printf("cls:%d;localangle=%f;\n",clsid, localangle);
|
||||
|
||||
if (localangle < 0 || localangle >= 90) {
|
||||
sigmaAmp[idx] = 0;
|
||||
}
|
||||
else {}
|
||||
|
||||
if (abs(tempsigma.p1)< PRECISIONTOLERANCE&&
|
||||
abs(tempsigma.p2) < PRECISIONTOLERANCE &&
|
||||
abs(tempsigma.p3) < PRECISIONTOLERANCE &&
|
||||
abs(tempsigma.p4) < PRECISIONTOLERANCE&&
|
||||
abs(tempsigma.p5) < PRECISIONTOLERANCE&&
|
||||
abs(tempsigma.p6) < PRECISIONTOLERANCE
|
||||
) {
|
||||
sigmaAmp[idx] = 0;
|
||||
}
|
||||
else {
|
||||
float sigma = GPU_getSigma0dB(tempsigma, localangle);
|
||||
sigma = powf(10.0, sigma / 10.0);// 后向散射系数
|
||||
//printf("cls:%d;localangle=%f;sigma0=%f;\n", clsid, localangle, sigma);
|
||||
sigmaAmp[idx] = sigma;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
|
||||
extern "C" void SatelliteAntDirectNormal(float* RstX, float* RstY, float* RstZ,
|
||||
float antXaxisX, float antXaxisY, float antXaxisZ,
|
||||
float antYaxisX, float antYaxisY, float antYaxisZ,
|
||||
float antZaxisX, float antZaxisY, float antZaxisZ,
|
||||
float antDirectX, float antDirectY, float antDirectZ,
|
||||
float* thetaAnt, float* phiAnt
|
||||
, long len) {
|
||||
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_SatelliteAntDirectNormal << <numBlocks, blockSize >> > (RstX, RstY, RstZ,
|
||||
antXaxisX, antXaxisY, antXaxisZ,
|
||||
antYaxisX, antYaxisY, antYaxisZ,
|
||||
antZaxisX, antZaxisY, antZaxisZ,
|
||||
antDirectX, antDirectY, antDirectZ,
|
||||
thetaAnt, phiAnt
|
||||
, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
|
||||
}
|
||||
|
||||
extern "C" void AntPatternInterpGain(float* anttheta, float* antphi, float* gain,
|
||||
float* antpattern, float starttheta, float startphi, float dtheta, float dphi, int thetapoints, int phipoints, long len) {
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
//printf("\nCUDA_RTPC_SiglePRF blockSize:%d ,numBlock:%d\n", blockSize, numBlocks);
|
||||
|
||||
CUDA_AntPatternInterpGain << <numBlocks, blockSize >> > ( anttheta,antphi, gain,
|
||||
antpattern,
|
||||
starttheta, startphi, dtheta, dphi, thetapoints, phipoints,
|
||||
len);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
|
||||
extern "C" void calculationEcho(float* sigma0, float* TransAnt, float* ReciveAnt,
|
||||
float* localangle, float* R, float* slopeangle,
|
||||
float nearRange, float Fs, float pt, float lamda, long FreqIDmax,
|
||||
cuComplex* echoAmp, long* FreqID,
|
||||
long len)
|
||||
{
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_calculationEcho << <numBlocks, blockSize >> > (sigma0, TransAnt, ReciveAnt,
|
||||
localangle, R, slopeangle,
|
||||
nearRange, Fs, pt, lamda, FreqIDmax,
|
||||
echoAmp, FreqID,
|
||||
len);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
|
||||
extern "C" void CUDAInterpSigma(
|
||||
long* demcls,float* sigmaAmp, float* localanglearr,long len,
|
||||
CUDASigmaParam* sigma0Paramslist, long sigmaparamslistlen) {// 地表覆盖类型-sigma插值对应函数-ulaby
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_InterpSigma << <numBlocks, blockSize >> > (
|
||||
demcls, sigmaAmp, localanglearr, len,
|
||||
sigma0Paramslist, sigmaparamslistlen
|
||||
);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
|
||||
#endif
|
||||
|
||||
|
||||
|
|
@ -0,0 +1,47 @@
|
|||
#ifndef _GPURTPC_H_
|
||||
#define _GPURTPC_H_
|
||||
|
||||
#include "BaseConstVariable.h"
|
||||
#include "GPUTool.cuh"
|
||||
#include <cuda_runtime.h>
|
||||
#include <device_launch_parameters.h>
|
||||
#include <cublas_v2.h>
|
||||
#include <cuComplex.h>
|
||||
|
||||
|
||||
extern "C" struct CUDASigmaParam {
|
||||
float p1;
|
||||
float p2;
|
||||
float p3;
|
||||
float p4;
|
||||
float p5;
|
||||
float p6;
|
||||
};
|
||||
|
||||
extern "C" void SatelliteAntDirectNormal(float* RstX, float* RstY, float* RstZ,
|
||||
float antXaxisX, float antXaxisY, float antXaxisZ,
|
||||
float antYaxisX, float antYaxisY, float antYaxisZ,
|
||||
float antZaxisX, float antZaxisY, float antZaxisZ,
|
||||
float antDirectX, float antDirectY, float antDirectZ,
|
||||
float* thetaAnt, float* phiAnt, long len);
|
||||
|
||||
extern "C" void AntPatternInterpGain(float* anttheta, float* antphi, float* gain,
|
||||
float* antpattern,
|
||||
float starttheta, float startphi, float dtheta, float dphi, int thetapoints, int phipoints,
|
||||
long len);
|
||||
|
||||
extern "C" void calculationEcho(float* sigma0, float* TransAnt, float* ReciveAnt,
|
||||
float* localangle, float* R, float* slopeangle,
|
||||
float nearRange, float Fs, float pt, float lamda, long FreqIDmax,
|
||||
cuComplex* echoAmp, long* FreqID,
|
||||
long len);
|
||||
|
||||
|
||||
extern "C" void CUDAInterpSigma(
|
||||
long* demcls, float* sigmaAmp, float* localanglearr, long len,
|
||||
CUDASigmaParam* sigma0Paramslist, long sigmaparamslistlen);
|
||||
|
||||
#endif
|
||||
|
||||
|
||||
|
||||
|
|
@ -0,0 +1,80 @@
|
|||
|
||||
|
||||
#include <iostream>
|
||||
#include <memory>
|
||||
#include <cmath>
|
||||
#include <complex>
|
||||
#include <device_launch_parameters.h>
|
||||
#include <cuda_runtime.h>
|
||||
#include <cublas_v2.h>
|
||||
#include <cuComplex.h>
|
||||
|
||||
#include "BaseConstVariable.h"
|
||||
#include "GPUTool.cuh"
|
||||
#include "GPUTBPImage.cuh"
|
||||
|
||||
#ifdef __CUDANVCC___
|
||||
|
||||
|
||||
|
||||
__global__ void CUDA_TBPImage(
|
||||
float* antPx, float* antPy, float* antPz,
|
||||
float* imgx, float* imgy, float* imgz,
|
||||
cuComplex* echoArr, cuComplex* imgArr,
|
||||
float freq, float fs, float Rnear, float Rfar,
|
||||
long rowcount, long colcount,
|
||||
long prfid, long freqcount
|
||||
) {
|
||||
int idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
//printf("\nidx:\t %d %d %d\n", idx, linecount, plusepoint);
|
||||
if (idx < rowcount * colcount) {
|
||||
float R = sqrtf(powf(antPx[prfid] - imgx[idx], 2) + powf(antPy[prfid] - imgy[idx], 2) + powf(antPz[prfid] - imgz[idx], 2));
|
||||
float Ridf = ((R - Rnear) * 2 / LIGHTSPEED) * fs;
|
||||
long Rid = floorf(Ridf);
|
||||
if(Rid <0|| Rid >= freqcount){}
|
||||
else {
|
||||
float factorj = freq * 4 * PI / LIGHTSPEED;
|
||||
cuComplex Rphi =cuCexpf(make_cuComplex(0, factorj * R));// 校正项
|
||||
imgArr[idx] = cuCaddf(imgArr[idx], cuCmulf(echoArr[Rid] , Rphi));// 矫正
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
extern "C" void CUDATBPImage(float* antPx, float* antPy, float* antPz,
|
||||
float* imgx, float* imgy, float* imgz,
|
||||
cuComplex* echoArr, cuComplex* imgArr,
|
||||
float freq, float fs, float Rnear, float Rfar,
|
||||
long rowcount, long colcount,
|
||||
long prfid, long freqcount)
|
||||
{
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (rowcount * colcount + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
//printf("\nCUDA_RTPC_SiglePRF blockSize:%d ,numBlock:%d\n",blockSize,numBlocks);
|
||||
// 调用 CUDA 核函数 CUDA_RTPC_Kernel
|
||||
|
||||
CUDA_TBPImage << <numBlocks, blockSize >> > (
|
||||
antPx, antPy, antPz,
|
||||
imgx, imgy, imgz,
|
||||
echoArr, imgArr,
|
||||
freq, fs, Rnear, Rfar,
|
||||
rowcount, colcount,
|
||||
prfid, freqcount
|
||||
);
|
||||
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDATBPImage CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
|
||||
|
||||
#endif
|
||||
|
||||
|
||||
|
|
@ -0,0 +1,37 @@
|
|||
#ifndef _GPUTBPIMAGE_H_
|
||||
#define _GPUTBPIMAGE_H_
|
||||
|
||||
#include "BaseConstVariable.h"
|
||||
#include "GPUTool.cuh"
|
||||
#include <cuda_runtime.h>
|
||||
#include <device_launch_parameters.h>
|
||||
#include <cublas_v2.h>
|
||||
#include <cuComplex.h>
|
||||
#include "GPUTool.cuh"
|
||||
|
||||
|
||||
|
||||
extern __global__ void CUDA_TBPImage(
|
||||
float* antPx, float* antPy, float* antPz,
|
||||
float* imgx, float* imgy, float* imgz,
|
||||
cuComplex* echoArr, cuComplex* imgArr,
|
||||
float freq, float fs, float Rnear, float Rfar,
|
||||
long rowcount, long colcount,
|
||||
long prfid, long freqcount
|
||||
);
|
||||
|
||||
extern "C" void CUDATBPImage(
|
||||
float* antPx,
|
||||
float* antPy,
|
||||
float* antPz,
|
||||
float* imgx,
|
||||
float* imgy,
|
||||
float* imgz,
|
||||
cuComplex* echoArr,
|
||||
cuComplex* imgArr,
|
||||
float freq, float fs, float Rnear, float Rfar,
|
||||
long rowcount, long colcount,
|
||||
long prfid, long freqcount
|
||||
);
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,274 @@
|
|||
|
||||
|
||||
#include <iostream>
|
||||
#include <memory>
|
||||
#include <cmath>
|
||||
#include <complex>
|
||||
#include <device_launch_parameters.h>
|
||||
#include <cuda_runtime.h>
|
||||
#include <cublas_v2.h>
|
||||
#include <cuComplex.h>
|
||||
|
||||
#include "BaseConstVariable.h"
|
||||
#include "GPUTool.cuh"
|
||||
|
||||
#ifdef __CUDANVCC___
|
||||
|
||||
|
||||
// 定义参数
|
||||
__device__ cuComplex cuCexpf(cuComplex x)
|
||||
{
|
||||
float factor = exp(x.x);
|
||||
return make_cuComplex(factor * cos(x.y), factor * sin(x.y));
|
||||
}
|
||||
|
||||
__device__ CUDAVector GPU_VectorAB(CUDAVector A, CUDAVector B) {
|
||||
CUDAVector C;
|
||||
C.x = B.x - A.x;
|
||||
C.y = B.y - A.y;
|
||||
C.z = B.z - A.z;
|
||||
return C;
|
||||
}
|
||||
|
||||
__device__ float GPU_VectorNorm2(CUDAVector A) {
|
||||
return sqrtf(A.x * A.x + A.y * A.y + A.z * A.z);
|
||||
}
|
||||
|
||||
__device__ float GPU_dotVector(CUDAVector A, CUDAVector B) {
|
||||
return A.x * B.x + A.y * B.y + A.z * B.z;
|
||||
}
|
||||
|
||||
__device__ float GPU_CosAngle_VectorA_VectorB(CUDAVector A, CUDAVector B) {
|
||||
return GPU_dotVector(A, B) / (GPU_VectorNorm2(A) * GPU_VectorNorm2(B));
|
||||
}
|
||||
|
||||
|
||||
__global__ void CUDA_DistanceAB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* R, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
R[idx] = sqrtf(powf(Ax[idx] - Bx[idx], 2) + powf(Ay[idx] - By[idx], 2) + powf(Az[idx] - Bz[idx], 2));
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_B_DistanceA(float* Ax, float* Ay, float* Az, float Bx, float By, float Bz, float* R, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
R[idx] = sqrtf(powf(Ax[idx] - Bx, 2) + powf(Ay[idx] - By, 2) + powf(Az[idx] - Bz, 2));
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_make_VectorA_B(float sX, float sY, float sZ, float* tX, float* tY, float* tZ, float* RstX, float* RstY, float* RstZ, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
RstX[idx] = sX - tX[idx]; // 地面->天
|
||||
RstY[idx] = sY - tY[idx];
|
||||
RstZ[idx] = sZ - tZ[idx];
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_Norm_Vector(float* Vx, float* Vy, float* Vz, float* R, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
R[idx] = sqrtf(powf(Vx[idx], 2) + powf(Vy[idx], 2) + powf(Vz[idx], 2));
|
||||
}
|
||||
}
|
||||
|
||||
__global__ void CUDA_cosAngle_VA_AB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* anglecos, long len) {
|
||||
long idx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
if (idx < len) {
|
||||
float tAx = Ax[idx];
|
||||
float tAy = Ay[idx];
|
||||
float tAz = Az[idx];
|
||||
float tBx = Bx[idx];
|
||||
float tBy = By[idx];
|
||||
float tBz = Bz[idx];
|
||||
float AR = sqrtf(powf(tAx, 2) + powf(tAy, 2) + powf(tAz, 2));
|
||||
float BR = sqrtf(powf(tBx, 2) + powf(tBy, 2) + powf(tBz, 2));
|
||||
float dotAB = tAx * tBx + tAy * tBy + tAz * tBz;
|
||||
float result = acosf(dotAB / (AR * BR));
|
||||
anglecos[idx] = result;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
//错误提示
|
||||
extern "C" void checkCudaError(cudaError_t err, const char* msg) {
|
||||
if (err != cudaSuccess) {
|
||||
std::cerr << "CUDA error: " << msg << " (" << cudaGetErrorString(err) << ")" << std::endl;
|
||||
exit(EXIT_FAILURE);
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
// 主机参数内存声明
|
||||
extern "C" void* mallocCUDAHost(long memsize) {
|
||||
void* ptr;
|
||||
cudaMallocHost(&ptr, memsize);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("mallocCUDAHost CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
return ptr;
|
||||
}
|
||||
|
||||
// 主机参数内存释放
|
||||
extern "C" void FreeCUDAHost(void* ptr) {
|
||||
cudaFreeHost(ptr);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("FreeCUDAHost CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
// GPU参数内存声明
|
||||
extern "C" void* mallocCUDADevice(long memsize) {
|
||||
void* ptr;
|
||||
cudaMalloc(&ptr, memsize);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("mallocCUDADevice CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
return ptr;
|
||||
}
|
||||
|
||||
// GPU参数内存释放
|
||||
extern "C" void FreeCUDADevice(void* ptr) {
|
||||
cudaFree(ptr);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("FreeCUDADevice CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
// GPU 内存数据转移
|
||||
extern "C" void HostToDevice(void* hostptr, void* deviceptr, long memsize) {
|
||||
cudaMemcpy(deviceptr, hostptr, memsize, cudaMemcpyHostToDevice);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("HostToDevice CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void DeviceToHost(void* hostptr, void* deviceptr, long memsize) {
|
||||
cudaMemcpy(hostptr, deviceptr, memsize, cudaMemcpyDeviceToHost);
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("DeviceToHost CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
// 基础运算函数
|
||||
extern "C" void CUDAdistanceAB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* R, long len) {
|
||||
// 设置 CUDA 核函数的网格和块的尺寸
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_DistanceAB << <numBlocks, blockSize >> > (Ax, Ay, Az, Bx, By, Bz, R, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void CUDABdistanceAs(float* Ax, float* Ay, float* Az, float Bx, float By, float Bz, float* R, long len) {
|
||||
// 设置 CUDA 核函数的网格和块的尺寸
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_B_DistanceA << <numBlocks, blockSize >> > (Ax, Ay, Az, Bx, By, Bz, R, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void CUDAmake_VectorA_B(float sX, float sY, float sZ, float* tX, float* tY, float* tZ, float* RstX, float* RstY, float* RstZ, long len) {
|
||||
// 设置 CUDA 核函数的网格和块的尺寸
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_make_VectorA_B << <numBlocks, blockSize >> > (sX, sY, sZ, tX, tY, tZ, RstX, RstY, RstZ, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void CUDANorm_Vector(float* Vx, float* Vy, float* Vz, float* R, long len) {
|
||||
// 设置 CUDA 核函数的网格和块的尺寸
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_Norm_Vector << <numBlocks, blockSize >> > (Vx, Vy, Vz, R, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
extern "C" void CUDAcosAngle_VA_AB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* anglecos, long len) {
|
||||
int blockSize = 256; // 每个块的线程数
|
||||
int numBlocks = (len + blockSize - 1) / blockSize; // 根据 pixelcount 计算网格大小
|
||||
// 调用 CUDA 核函数
|
||||
CUDA_cosAngle_VA_AB << <numBlocks, blockSize >> > (Ax, Ay, Az, Bx, By, Bz, anglecos, len);
|
||||
|
||||
#ifdef __CUDADEBUG__
|
||||
cudaError_t err = cudaGetLastError();
|
||||
if (err != cudaSuccess) {
|
||||
printf("CUDA_RTPC_SiglePRF CUDA Error: %s\n", cudaGetErrorString(err));
|
||||
// Possibly: exit(-1) if program cannot continue....
|
||||
}
|
||||
#endif // __CUDADEBUG__
|
||||
cudaDeviceSynchronize();
|
||||
}
|
||||
|
||||
#endif
|
||||
|
||||
|
||||
|
|
@ -0,0 +1,74 @@
|
|||
#ifndef _GPUTOOL_H_
|
||||
#define _GPUTOOL_H_
|
||||
#ifdef __CUDANVCC___
|
||||
#include "BaseConstVariable.h"
|
||||
#include <cuda_runtime.h>
|
||||
#include <device_launch_parameters.h>
|
||||
#include <cublas_v2.h>
|
||||
#include <cuComplex.h>
|
||||
|
||||
#define __CUDADEBUG__
|
||||
|
||||
#define CUDAMEMORY Memory1MB*100
|
||||
|
||||
#define LAMP_CUDA_PI 3.141592653589793238462643383279
|
||||
|
||||
|
||||
// 默认显存分布
|
||||
|
||||
|
||||
enum LAMPGPUDATETYPE {
|
||||
LAMP_LONG,
|
||||
LAMP_FLOAT,
|
||||
LAMP_COMPLEXFLOAT
|
||||
};
|
||||
|
||||
extern "C" struct CUDAVector {
|
||||
float x;
|
||||
float y;
|
||||
float z;
|
||||
};
|
||||
|
||||
extern "C" struct CUDAVectorEllipsoidal {
|
||||
float theta;
|
||||
float phi;
|
||||
float pho;
|
||||
};
|
||||
|
||||
// 定义设备函数
|
||||
extern __device__ cuComplex cuCexpf(cuComplex x);
|
||||
extern __device__ CUDAVector GPU_VectorAB(CUDAVector A, CUDAVector B);
|
||||
extern __device__ float GPU_VectorNorm2(CUDAVector A);
|
||||
extern __device__ float GPU_dotVector(CUDAVector A, CUDAVector B);
|
||||
extern __device__ float GPU_CosAngle_VectorA_VectorB(CUDAVector A, CUDAVector B);
|
||||
|
||||
|
||||
// 定义全局函数
|
||||
extern __global__ void CUDA_DistanceAB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* R, long len);
|
||||
extern __global__ void CUDA_B_DistanceA(float* Ax, float* Ay, float* Az, float Bx, float By, float Bz, float* R, long len);
|
||||
extern __global__ void CUDA_make_VectorA_B(float sX, float sY, float sZ, float* tX, float* tY, float* tZ, float* RstX, float* RstY, float* RstZ, long len);
|
||||
extern __global__ void CUDA_Norm_Vector(float* Vx, float* Vy, float* Vz, float* R, long len);
|
||||
extern __global__ void CUDA_cosAngle_VA_AB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* anglecos, long len);
|
||||
|
||||
|
||||
// 误差处理函数
|
||||
extern "C" void checkCudaError(cudaError_t err, const char* msg);
|
||||
|
||||
// GPU 内存函数
|
||||
extern "C" void* mallocCUDAHost( long memsize); // 主机内存声明
|
||||
extern "C" void FreeCUDAHost(void* ptr);
|
||||
extern "C" void* mallocCUDADevice( long memsize); // GPU内存声明
|
||||
extern "C" void FreeCUDADevice(void* ptr);
|
||||
extern "C" void HostToDevice(void* hostptr, void* deviceptr, long memsize);//GPU 内存数据转移 设备 -> GPU
|
||||
extern "C" void DeviceToHost(void* hostptr, void* deviceptr, long memsize);//GPU 内存数据转移 GPU -> 设备
|
||||
|
||||
|
||||
// 基础运算函数
|
||||
extern "C" void CUDAdistanceAB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* R, long member);
|
||||
extern "C" void CUDABdistanceAs(float* Ax, float* Ay, float* Az, float Bx, float By, float Bz, float* R, long member);
|
||||
extern "C" void CUDAmake_VectorA_B(float sX, float sY, float sZ, float* tX, float* tY, float* tZ, float* RstX, float* RstY, float* RstZ, long member);
|
||||
extern "C" void CUDANorm_Vector(float* Vx, float* Vy, float* Vz, float* R, long member);
|
||||
extern "C" void CUDAcosAngle_VA_AB(float* Ax, float* Ay, float* Az, float* Bx, float* By, float* Bz, float* anglecos, long len);
|
||||
|
||||
#endif
|
||||
#endif
|
||||
|
|
@ -68,7 +68,7 @@
|
|||
<IncludePath>.\SimulationSAR;.\GF3ProcessToolbox;.\BaseTool;$(IncludePath)</IncludePath>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)|$(Platform)' == 'Release|x64'">
|
||||
<IncludePath>.\SimulationSAR;.\GF3ProcessToolbox;.\BaseTool;$(oneMKLIncludeDir);$(IncludePath)</IncludePath>
|
||||
<IncludePath>.\GPUTool;.\SimulationSAR;.\GF3ProcessToolbox;.\BaseTool;$(oneMKLIncludeDir);$(IncludePath)</IncludePath>
|
||||
</PropertyGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
|
||||
<ClCompile>
|
||||
|
|
@ -184,13 +184,15 @@
|
|||
<ClInclude Include="BaseTool\EchoDataFormat.h" />
|
||||
<ClInclude Include="BaseTool\FileOperator.h" />
|
||||
<ClInclude Include="BaseTool\GeoOperator.h" />
|
||||
<ClInclude Include="BaseTool\GPUTool.cuh" />
|
||||
<ClInclude Include="BaseTool\ImageOperatorBase.h" />
|
||||
<ClInclude Include="BaseTool\LogInfoCls.h" />
|
||||
<QtMoc Include="BaseTool\QToolProcessBarDialog.h" />
|
||||
<ClInclude Include="BaseTool\RasterToolBase.h" />
|
||||
<ClInclude Include="BaseTool\SARSimulationImageL1.h" />
|
||||
<ClInclude Include="BaseTool\stdafx.h" />
|
||||
<ClInclude Include="GPUTool\GPURTPC.cuh" />
|
||||
<ClInclude Include="GPUTool\GPUTBPImage.cuh" />
|
||||
<ClInclude Include="GPUTool\GPUTool.cuh" />
|
||||
<ClInclude Include="SimulationSAR\TBPImageAlgCls.h" />
|
||||
<QtMoc Include="QSimulationRTPCGUI.h" />
|
||||
<QtMoc Include="GF3ProcessToolbox\QOrthSlrRaster.h" />
|
||||
|
|
@ -213,10 +215,18 @@
|
|||
<QtMoc Include="QMergeRasterProcessDialog.h" />
|
||||
</ItemGroup>
|
||||
<ItemGroup>
|
||||
<CudaCompile Include="BaseTool\GPUTool.cu" />
|
||||
<None Include="cpp.hint" />
|
||||
</ItemGroup>
|
||||
<ItemGroup>
|
||||
<None Include="cpp.hint" />
|
||||
<CudaCompile Include="GPUTool\GPURTPC.cu">
|
||||
<GenerateRelocatableDeviceCode Condition="'$(Configuration)|$(Platform)'=='Release|x64'">true</GenerateRelocatableDeviceCode>
|
||||
</CudaCompile>
|
||||
<CudaCompile Include="GPUTool\GPUTBPImage.cu">
|
||||
<GenerateRelocatableDeviceCode Condition="'$(Configuration)|$(Platform)'=='Release|x64'">true</GenerateRelocatableDeviceCode>
|
||||
</CudaCompile>
|
||||
<CudaCompile Include="GPUTool\GPUTool.cu">
|
||||
<GenerateRelocatableDeviceCode Condition="'$(Configuration)|$(Platform)'=='Release|x64'">true</GenerateRelocatableDeviceCode>
|
||||
</CudaCompile>
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Condition="Exists('$(QtMsBuild)\qt.targets')">
|
||||
|
|
|
|||
|
|
@ -21,6 +21,7 @@
|
|||
|
||||
#ifdef __CUDANVCC___
|
||||
#include "GPUTool.cuh"
|
||||
#include "GPURTPC.cuh"
|
||||
#endif // __CUDANVCC___
|
||||
#include <Imageshow/ImageShowDialogClass.h>
|
||||
|
||||
|
|
@ -827,9 +828,9 @@ ErrorCode RTPCProcessCls::RTPCMainProcess_GPU( )
|
|||
std::cout << "ant Position=[" << antpx << "," << antpy << "," << antpz << "]" << std::endl;
|
||||
#endif // __PRFDEBUG__
|
||||
|
||||
make_VectorA_B(antpx, antpy, antpz, d_dem_x, d_dem_y, d_dem_z, d_RstX, d_RstY, d_RstZ, pixelcount); // Rst = Rs - Rt; 華醱-> 硌砃
|
||||
Norm_Vector(d_RstX, d_RstY, d_RstZ, d_R, pixelcount); // R
|
||||
cosAngle_VA_AB(d_RstX, d_RstY, d_RstZ, d_demsloper_x, d_demsloper_y, d_demsloper_z, d_localangle, pixelcount); // 擁窒⻌扞褒
|
||||
CUDAmake_VectorA_B(antpx, antpy, antpz, d_dem_x, d_dem_y, d_dem_z, d_RstX, d_RstY, d_RstZ, pixelcount); // Rst = Rs - Rt; 華醱-> 硌砃
|
||||
CUDANorm_Vector(d_RstX, d_RstY, d_RstZ, d_R, pixelcount); // R
|
||||
CUDAcosAngle_VA_AB(d_RstX, d_RstY, d_RstZ, d_demsloper_x, d_demsloper_y, d_demsloper_z, d_localangle, pixelcount); // 擁窒⻌扞褒
|
||||
SatelliteAntDirectNormal(d_RstX, d_RstY, d_RstZ,
|
||||
antXaxisX, antXaxisY, antXaxisZ,
|
||||
antYaxisX, antYaxisY, antYaxisZ,
|
||||
|
|
|
|||
|
|
@ -7,6 +7,7 @@
|
|||
#include <QProgressDialog>
|
||||
#include <QMessageBox>
|
||||
#include "GPUTool.cuh"
|
||||
#include "GPUTBPImage.cuh"
|
||||
|
||||
void CreatePixelXYZ(std::shared_ptr<EchoL0Dataset> echoL0ds, QString outPixelXYZPath)
|
||||
{
|
||||
|
|
@ -400,9 +401,6 @@ void TBPImageGPUAlg(std::shared_ptr<float> antPx, std::shared_ptr<float> antPy,
|
|||
FreeCUDADevice(d_echoArr);
|
||||
FreeCUDADevice(d_imgArr);
|
||||
|
||||
|
||||
// ΚΝ·ΕGPU±δΑΏ
|
||||
|
||||
}
|
||||
|
||||
|
||||
|
|
@ -464,8 +462,6 @@ void TBPImageGPUAlg(std::shared_ptr<float> antPx, std::shared_ptr<float> antPy,
|
|||
|
||||
|
||||
/**
|
||||
|
||||
|
||||
ErrorCode TBPImageAlgCls::ProcessCPU(long num_thread)
|
||||
{
|
||||
omp_set_num_threads(num_thread);
|
||||
|
|
|
|||
Loading…
Reference in New Issue