Feature matching using Euclidean

Question

Im 2015-5-13

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/216482-feature-matching-using-euclidean

评论： Im 2015-5-18

I am now doing copy move image forgery detection using DCT. Now I have some problem for matching part. It lagging for loading whole data since I am using looping to get the similarity between two row vector. Any idea to change my code in better way. Here is my code for the matching part.

z=0;
for num=1:group    
    for a=mS(num):mE(num)-1
        for b=a+1:mE(num)
            z=z+1;
            Dsimilar(z,1) = norm(sort(a,:)-sort(b,:))
        end
    end
end

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Walter Roberson 2015-5-13

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/216482-feature-matching-using-euclidean#answer_178906

That is not a valid calling form for the sort() function, so that code will not execute.

Why would you want to sort a single integer anyhow?

3 个评论
显示 1更早的评论隐藏 1更早的评论

Walter Roberson 2015-5-13

在 MATLAB Online 中打开

norm() by default calculates the 2-norm, which is the Euclidean length of the vector. Which is sqrt(sum(x.^2)) . You can calculate a lot of euclidean lengths at once. For example,

for a = Ms(num):Me(num)+1
  Ta = repmat(sortZZ(a,:), Me(num)-a, 1);
  Tb = sortZZ(a+1:me(num),:);
  norms = sqrt(sum((Ta - Tb).^2, 2));
  Dsimilar = [Dsimilar; norms(:)];
end