CUDA+OPENCV混合编程之访问GpuMat中的每一个像素

一、环境配置

Ubuntu + CUDA + OPENCV。先安装NVIDIA显卡驱动，再安装CUDA，最后编译安装OpenCV(加上了CUDA的模块)，具体怎么做有很多教程，我就不多说了。

二、代码

//main.cpp#include <iostream>#include "kernel.h"#define WIDTH 1024#define HEIGHT 512int main(){    GpuMat mat(HEIGHT,WIDTH,CV_8UC3);    generateFrameCaller(WIDTH,HEIGHT,mat);    HANDLECUDAERROR(cudaGetLastError()); //Check error after call kernel function    Mat dst;    mat.download(dst);    imshow("result",dst);    waitKey(0);    return 0;}

//kernel.h#ifndef KERNEL_H_INCLUDED#define KERNEL_H_INCLUDED#include <cuda.h>#include <cuda_runtime.h>#include <opencv2/opencv.hpp>#include <opencv2/gpu/gpu.hpp>#include <opencv2/gpu/gpu.hpp>#include <opencv2/core/cuda_devptrs.hpp>#include <opencv2/gpu/stream_accessor.hpp>using namespace cv;using namespace cv::gpu;void generateFrameCaller(int width,int height,PtrStep<uchar3> mat);#endif // KERNEL_H_INCLUDED

//kernel.cu#include "kernel.h"/** *　访问图片中的每一个像素 * @param width mat对应的宽度 * @param height mat对应的高度 * @param mat GpuMat被强转为PtrStep<>类型之后的参数 */__global__ void generateFrame(int width,int height,PtrStep<uchar3> mat){    int x = threadIdx.x + blockIdx.x * blockDim.x;    int y = threadIdx.y + blockIdx.y * blockDim.y;    if(x < width && y < height)    {        mat(y,x) = make_uchar3(255,123,123);     //mat(y,x)才是正确打开方式    }}/** *　调用内核函数的主机函数 * @param width mat对应的宽度 * @param height mat对应的高度 * @param mat GpuMat被强转为PtrStep<>类型之后的参数 */void generateFrameCaller(int width,int height,PtrStep<uchar3> mat){    dim3 blocks(width/32,height/32);    dim3 threads(32,32);    generateFrame<<<blocks,threads>>>(width,height,mat);}

# CMakeLists.txt# required cmake versioncmake_minimum_required(VERSION 2.8)# CMakeLists.txt for G4CU projectproject(cudatest)# packagesfind_package(CUDA REQUIRED)find_package(Ceres REQUIRED)#set(OpenCV_DIR "/usr/local/include")       #修改了这一句，指定opencv头文件的路径find_package(OpenCV REQUIRED)set(CMAKE_BUILD_TYPE "RELEASE")set(CUDA_USE_STATIC_CUDA_RUNTIME OFF)       #这一句解决 cannot find -lopencv_dep_cudart# nvcc flags#set(CUDA_NVCC_FLAGS -gencode arch=compute_20,code=sm_20;-G;-g)set(CUDA_NVCC_FLAGS -gencode arch=compute_52,code=sm_52;-G;-g)include_directories( ${OPENCV_INCLUDE_DIRS})include_directories( ${CERES_INCLUDE_DIRS})include_directories(include)file(GLOB_RECURSE CURRENT_HEADERS  include/*.h include/*.hpp include/*.cuh)file(GLOB CURRENT_SOURCES  src/*.c src/*.cpp src/*.cu)source_group("Include" FILES ${CURRENT_HEADERS}) source_group("Source" FILES ${CURRENT_SOURCES}) CUDA_ADD_EXECUTABLE(cudatest ${CURRENT_HEADERS} ${CURRENT_SOURCES})target_link_libraries( ${PROJECT_NAME} ${OpenCV_LIBS} ${CERES_LIBRARIES})

三、可能遇到的Bug

解决cannot find -lopencv_dep_cudart问题