c++cimg

How to remove pepper noise using median filter algorithm


I have written a code (in c++) to add noise in the image using CImg library. Now i want to load the image with noise and remove those noise inside the image using median filter algorithm. Below is my code.

int main()
{
    int x;
    cout<<"Welcome to my app\n";
    cout<<"Choose options below\n";
    cout<<"1. Remove pepper    2. Add pepper\n";
    cin>>x;
    if (x==1)
    {
        cout<<"Needs help";
        /* 
        * i tried to change the noise level to 0 but it did not work like below 
        * image.noise(0,2); 
        * 
        */
    }
    else if(x==2)
    {
        //image file
        CImg<unsigned char> image("new.bmp");
        const unsigned char red[] = { 255,0,0 }, green[] = { 0,255,0 }, blue[] = { 0,0,255 };
        image.noise(100,2);
        image.save("new2.bmp");
        CImgDisplay main_disp(image, "Image with Pepper noise");
        while (!main_disp.is_closed())
            {
                main_disp.wait();               
            }
    }

    getchar();
    return 0;

}

If there is another way of doing this using CImg library, I will be thankful!!


Solution

  • According to this tutorial and the definition of the median function you will get something like this:

    #include <algorithm>
    using namespace std;
    // ...
    int ksize = 3; // 5, 7, N and so on... for NxN kernel
    int ksize2 = ksize/2;
    vector<uchar> kernel(ksize*ksize, 0);
    for (int i=ksize2;i<image.dimx() - ksize2;i++)
        for (int j=ksize2;j<image.dimy() - ksize2;j++)
            for (int k=0;k<3;k++) {
                // prepare kernel
                int n = 0;
                for(int l = -ksize2; l <= ksize2; l++)
                    for(int m = -ksize2; m <= ksize2; m++)
                        kernel[n++] = image(i + l,j + m,0,k); 
    
                // using std::algorithm to find median
                sort(kernel.begin(), kernel.end());
    
                // simple assign median value to created empty image
                medianFilteredImage(i, j, 0, k) = kernel[kernel.size()/2]; // median is here now
            }
        }
    }