CUDA和C++函数问题(Visual Studio 2013）

我正试图从Visual Studio中的cpp文件调用cu文件中定义的cuda函数，但我一直收到以下错误。

TomColourCorrection Main.obj：错误LNK2019：未解析的外部符号"public:void __cdecl hwk:：TomColourCorrection：：bright（int，int）"（？brightness@TomColourCorrection@hwk@@QEAAXHH@Z)在函数"public:virtual void __cdecl hwk:：TomColourCorrection：：processCore（class std:：shared_ptr）"中引用（？processCore@TomColourCorrection@hwk@@UEAAXV$shared_ptr@VIImageProcessingContext@hwk@@@std@@@Z）

现在，通过阅读其他类似的问题，我理解了它与函数的定义有关，也理解了其中存在问题，但当我在头文件和cuda文件中定义时，我看不到它。

这是我的代码（我是CUDA的新手，但我可以很好地编译CUDA，当我不在C++中调用这个函数时，代码就会运行）：

头文件

#pragma once
#include "ImageProcessorWithProperties.h"
#include <iostream>
#include <cuda_runtime.h>
#include    <cuda.h>
class TomColourCorrection : public ImageProcessorWithProperties, public      PropertyConsumer<TomColourCorrection>{
public: TomColourCorrection(PropNodePtr n, std::function<void()> requestReprocess);
    virtual void processCore(IImageProcessingContextPtr context);
static void DeclareSettings(hwk::PropNodePtr n);
virtual ~TomColourCorrection();
void brightness(int iw, int ih); (function I am talking about)
};
 }

带有函数调用的cpp文件//它只是重要代码的片段，因为它的其余部分对于实际函数本身来说是不必要的

 #include "stdafx.h"
 #include "TomColourCorrection.h"
 #include <opencv2/imgproc/imgproc.hpp>
 #include <cv.h>
 #include <highgui.h>
 #include <opencv2/core/core.hpp>
 #include <opencv2/highgui/highgui.hpp>
 #include <iostream>
 #include <cuda_runtime.h>
 #include   <cuda.h>
 namespace hwk{ 
    TomColourCorrection::TomColourCorrection(PropNodePtr n, std::function<void()> requestReprocess) :
ImageProcessorWithProperties("sandbox", n, requestReprocess),
PropertyConsumer<TomColourCorrection>(n)
{
 }
void TomColourCorrection::processCore(IImageProcessingContextPtr context){

 brightness(16, 16); (just generic numbers at the moment as I am trying to resolve this issue etc)
    }
  }

CUDA文件和功能定义

#include "TomColourCorrection.h"
#include "device_launch_parameters.h"
__global__ void brightness_kernel(int iw, int ih)
{
// Calculate our pixel's location
int x = (blockIdx.x * blockDim.x) + threadIdx.x;
int y = (blockIdx.y * blockDim.y) + threadIdx.y;
// Variables to store the sum
int count = 0;
float sum = 0.0;
// Do the blur operation by summing the surround pixels
/*  for (int j = -(bh / 2); j <= (bh / 2); j++)
{
    for (int i = -(bw / 2); i <= (bw / 2); i++)
    {
        // Verify that this offset is within the image boundaries
        if ((x + i) < iw && (x + i) >= 0 && (y + j) < ih && (y + j) >= 0)
        {
            sum += (float)source[((y + j) * iw) + (x + i)];
            count++;
        }
    }
}*/
// Average the sum
sum /= (float)count;
//  dest[(y * iw) + x] = (unsigned char)sum;
}

void brightness(int iw, int ih) //, unsigned char *source, unsigned char *dest)
{
// allocate memory for the bitmap in GPU memory
unsigned char *dev_source, *dev_dest;
//  cudaHostGetDevicePointer(&dev_source, source, 0);
//  cudaHostGetDevicePointer(&dev_dest, dest, 0);
// Run the boxfilter kernel
dim3 blocks(iw / 16, ih / 16);
dim3 threads(16, 16);
// Execute the kernel
brightness_kernel << <blocks, threads >> >(iw, ih);
cudaThreadSynchronize();
}