OpenCV2.4.4+CUDA+VS2010实现图像的反转

    实现图像的反转并不难，小编在opencv篇中已经把代码贴出。本文却又旧事重提，主要为了用CUDA实现并行处理反转过程，代码难度并不大，但是却是对opencv+CUDA结合起来的一个应用，给以后一些运算量大的图像算法用CUDA并行运算实现提供一个现实参考，好了，废话不多说，先贴上代码：

/*****************************************************************************************************
 *    File name: cudaCppMain.cpp
 *    Author: source code by RamiYim
 *    Version: cs-hhu-cn0.0
 *    Date: 2014.08.12
 *    Description: 测试cuda与opencv相结合
 *    Others:
 *    Function List: main functions and its function
 *    History:<rami>,<20140819>,<添加相关函数>,<desc> 
      <rami>,<20140812>,<创建文件>,<desc>
******************************************************************************************************/
#include "reverseImage.h"

int main()
{
    reverseImage(cvLoadImage("D:/lena.jpg"));
    return 0;
}

/*****************************************************************************************************
 *    File name: reverseImage.h
 *    Author: source code by RamiYim
 *    Version: cs-hhu-cn0.0
 *    Date: 2014.08.12
 *    Description: 测试cuda与opencv相结合
 *    Others:
 *    Function List: main functions and its function
 *    History:<rami>,<20140819>,<添加相关函数>,<desc> 
      <rami>,<20140812>,<创建文件>,<desc>
******************************************************************************************************/
#include "cv.h"
#include "highgui.h"
void reverseImage(IplImage *);

/*****************************************************************************************************
 *    File name: reverseImage.cpp
 *    Author: source code by RamiYim
 *    Version: cs-hhu-cn0.0
 *    Date: 2014.08.12
 *    Description: 测试cuda与opencv相结合
 *    Others:
 *    Function List: 实现reverseImage函数
 *    History:<rami>,<20140819>,<添加相关函数>,<desc> 
 *            <rami>,<20140812>,<创建文件>,<desc>
******************************************************************************************************/
#include "reverseImage.h"
#include "cv.h"
extern "C" int revImg(const char *, int size);
void reverseImage(IplImage *sourceImage){ 
    cvNamedWindow("sourceImage"); 
    cvShowImage("sourceImage", sourceImage); 
    cvWaitKey(0); 
    int width, height, channels, width_step; 
    width = sourceImage -> width; 
    height = sourceImage -> height; 
    channels = sourceImage -> nChannels; 
    width_step = sourceImage -> widthStep; 
    const char *data = sourceImage -> imageData;
    int size = width * height * channels; 
    revImg(data, size); 
    cvShowImage("reversedImage", sourceImage);
    cvWaitKey(0);
    cvDestroyWindow("sourceImage"); 
    cvDestroyWindow("reversedImage");
}

#include "cuda_runtime.h"
#include "device_launch_parameters.h"
#include <stdio.h>
#include "cv.h"
#include "highgui.h"

cudaError_t addWithCuda(char *data, int size);

__global__ void revImg(char *dev_c)
{
    int bid = blockIdx.x * gridDim.x + blockIdx.y;   //gridDim.x为固定值400，blockIdx.x取值范围为0~399
    int tid = bid * blockDim.x + threadIdx.x;        //blockDim.x为定值3，threadIdx.x取值范围为0~2

    dev_c[tid] = 255 - dev_c[tid];                   //实现反转主体功能
}

extern "C" void revImg(char *data, int size)
{
    cudaError_t cudaStatus = addWithCuda(data, size);
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "addWithCuda failed!");
    }
}

cudaError_t addWithCuda(char *data, int size) {
    char *dev_c = 0;
    cudaError_t cudaStatus;

    // Choose which GPU to run on, change this on a multi-GPU system.
    cudaStatus = cudaSetDevice(0);
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "cudaSetDevice failed!  Do you have a CUDA-capable GPU installed?");
        goto Error;
    }

    // Allocate GPU buffers for dev_c
    cudaStatus = cudaMalloc((void**)&dev_c, size * sizeof(char));
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "cudaMalloc failed!");
        goto Error;
    }

    // Copy input vectors from host memory to GPU buffers.
    cudaStatus = cudaMemcpy(dev_c, data, size * sizeof(char), cudaMemcpyHostToDevice);
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "cudaMemcpy failed!");
        goto Error;
    }
	
    // Launch a kernel on the GPU with one thread for each element.
    dim3 dimGrid(400, 400);
    dim3 dimBlock(3, 1);        //1行3列
    revImg<<<dimGrid, dimBlock>>>(dev_c);        //调用内核函数

    // Check for any errors launching the kernel
    cudaStatus = cudaGetLastError();
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "addKernel launch failed: %s\n", cudaGetErrorString(cudaStatus));
        goto Error;
    }

    // cudaDeviceSynchronize waits for the kernel to finish, and returns
    // any errors encountered during the launch.
    cudaStatus = cudaDeviceSynchronize();
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "cudaDeviceSynchronize returned error code %d after launching addKernel!\n", cudaStatus);
        goto Error;
    }

    // Copy output vector from GPU buffer to host memory.
    cudaStatus = cudaMemcpy(data, dev_c, size * sizeof(char), cudaMemcpyDeviceToHost);
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "cudaMemcpy failed!");
        goto Error;
    }

Error:
    cudaFree(dev_c);

    return cudaStatus;
}

以上即为函数的主体功能，结果如下图：

PS:CUDA环境配置见本人博客，地址为：http://blog.youkuaiyun.com/a2112233445566/article/details/38420567

       OpenCV环境配置件本人博客，地址为：http://blog.youkuaiyun.com/a2112233445566/article/details/38540389