我想在Matlab中用RGB像素匹配这两个图像..我猜我的代码还可以..但它不起作用 [英] I want to match this two image by RGB pixel in Matlab..I guess my code is ok..but it's not working

问题描述

在这里，我使用的是主图像"cricketteam.jpg"，并裁剪了一部分名为"musfiq.jpg"的图像.但是从我的代码来看，即使它们相同，我什至无法匹配单个像素.有什么问题吗?

Here I am using main image ''cricketteam.jpg' and cropped a portion of it named 'musfiq.jpg'. But from my code, I can't even match a single pixel though they are same. What is the problem?

这是我的代码:

function imageMatch
    i1=imread('cricketteam.jpg');
    [y,x,d]=size(i1);
    m=imread('musfiq.jpg');
    [y1,x1,d]=size(m);
    x2=1;
    y2=1;
    row_match=0;
    flag_y=0;
    cx=1;
    xc=1;
    xc1=1;


s_loop=x-x1;

s_loop= s_loop+1;
for i=1:y
   cx=xc;

    for j=cx:x


        if j==s_loop && row_match==0
            break;
        end
        f=impixel(m,x2,y2);

        ma=impixel(i1,j,i);

        end
        if isequal(f,ma)==1

             if  row_match == 0
                xc=cx;
             end
            row_match=row_match+1;
            x2=x2+1;
        else

              flag_y=0;
              row_match=0;
              y2=1;
              x2=1;

        end
        if row_match==x1
              flag_y=flag_y+1;
              row_match=0;
              y2=y2+1;
              x2=1;
              break;
        end

    end

    if  flag_y==y1
        disp('Image Matched')
        break;
    end
end

end

这里是 cricketteam.jpg :

这是 musfiq.jpg 的部分:

推荐答案

解决方案

Jpeg图像包含jpeg压缩，因此您要查找的补丁不会在图像中原样显示. 相反，您可以查找最大化归一化互相关( NCC). NCC是一种评估两个补丁之间相似度的方法. 为此，请使用MATLAB的 normxcorr2 函数.

Jpeg images include jpeg compression, and therefore the patch which you are looking for is not going to appear as-is in the image. Instead, you can look for the location of the patch which maximizes the normalized cross correlation (NCC) between the images. The NCC is a measure which evaluates the level of similarity between two patches. In order to do so, use MATLAB's normxcorr2 function.

代码

%reads images
I = rgb2gray(imread('team.jpg')); 
template = rgb2gray(imread('p.jpg')); 
[h,w]=size(template);

%calculate the ncc at each pixel
c = normxcorr2(template,I);

%finds the maximal ncc pixel
[y, x] = find(c==max(c(:)));

%extracts the relevant patch
yoffSet = ypeak-size(template,1);
xoffSet = xpeak-size(template,2);
patch = I(y-h+1:y,x-w+1:x);

结果

数字结果:

C(y,x) = 0.9992 %ncc result (complete match is 1)
[y, x]= 84 342 %most suitable patch center

图像输出:

%display the original image with the found patch
hFig = figure;
hAx  = axes;
imshow(I,'Parent', hAx);
imrect(hAx, [xoffSet+1, yoffSet+1, size(template,2), size(template,1)]);

这篇关于我想在Matlab中用RGB像素匹配这两个图像..我猜我的代码还可以..但它不起作用的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！