compare.m

来自「用matlab进行图像处理 对车牌号码进行识别 非常具有实用价值」· M 代码 · 共 92 行

function [absdiff, snr, psnr, imfid, mse] = compare(originalimg, restoredimg)
%Function to compare the original image with the
%restored image using some predefined methods
%Inputs: originalimg, restoredimg
%Returns: absdiff, snr, psnr, imfid, mse
%
%originalimg:  It is the original unblurred image
%restoredimg:  It is the restored image
%absdiff:      Average Absolute Difference
%snr:          Signal to Noise Ratio (dB)
%psnr:         Peak Signal to Noise Ratio (dB)
%imfid:        Image Fidelity
%mse:          Mean Square Error
%
%Example:
%      [absdiff, snr, psnr, imfid, mse] = compare(originalimg, restoredimg);
%      This call takes original image and the restored image and returns
%      the comparison of both the images using some predefined methods.

%Average absolute difference
%{
md = originalimg - restoredimg;
mdsize = size(md);
summation = 0;
for i = 1:mdsize(1);
    for j = 1:mdsize(2);
        summation = summation + abs(md(i,j));
    end
end

absdiff = summation/(mdsize(1)*mdsize(2));
%}


%Signal to Noise Ratio (SNR)
%{
md = (originalimg - restoredimg).^2;
mdsize = size(md);
summation = 0;
sumsq=0;
for  i = 1:mdsize(1);
    for j = 1:mdsize(2);
        summation = summation + abs(md(i,j));
        sumsq = sumsq + (originalimg(i,j)^2);
    end
end

snr = sumsq/summation;
snr = 10 * log10(snr);
%}


%Peak Signal to Noise Ratio (PSNR)
%{
md = (originalimg - restoredimg).^2;
mdsize = size(md);
summation = 0;
sumsq=0;
for  i = 1:mdsize(1);
    for j = 1:mdsize(2);
        summation = summation + abs(md(i,j));
    end
end

psnr = size(originalimg, 1) * size(originalimg, 2) * max(max(originalimg.^2))/summation;
psnr = 10 * log10(psnr);
%}


%Image Fidelity

md = (originalimg - restoredimg).^2;
mdsize = size(md);
summation = 0;
sumsq = 0;
for  i = 1:mdsize(1);
    for j = 1:mdsize(2);
        summation = summation + abs(md(i,j));
        sumsq = sumsq + (originalimg(i,j)^2);
    end
end

imfid = (1-summation)/sumsq;
%}


%Mean Square Error
%{
diff = originalimg - restoredimg;
diff1 = diff.^2;
mse = mean(mean(diff1));
%}