GWL--SSIM代码_gmgwl-优快云博客

本文链接：https://blog.youkuaiyun.com/m0_62404686/article/details/131259508

该代码实现了一个函数GWL_SSIM，用于计算两幅图像（灰度或彩色）之间的相似度得分，主要考虑图像的亮度成分。通过平均滤波、梯度映射和变异数映射等步骤，计算结构相似度矩阵，最终得出GWL_SSIM值，用于表示图像质量。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

function GWL_SSIM = GWL_SSIM(imageRef, imageDis)
%Input : (1) imageRef: the first image being compared
%        (2) imageDis: the second image being compared
%
%Output: (1) GWL_SSIM: is the similarty score calculated using GWL_SSIM algorithm. GWL_SSIM
%	     only considers the luminance component of images.       
%-----------------------------------------------------------------------
%
%Usage:
%Given 2 test images img1 and img2. For gray-scale images, their dynamic range should be 0-255.
%For colorful images, the dynamic range of each color channel should be 0-255.
% GWL_SSIM = GWL_SSIM(img1, img2);
%-----------------------------------------------------------------------

[rows, cols, junk] = size(imageRef);
if junk == 3  %images are colorful
    Y1 = 0.299 * double(imageRef(:,:,1)) + 0.587 * double(imageRef(:,:,2)) + 0.114 * double(imageRef(:,:,3));
    Y2 = 0.299 * double(imageDis(:,:,1)) + 0.587 * double(imageDis(:,:,2)) + 0.114 * double(imageDis(:,:,3));
else %images are grayscale
    Y1 = double(imageRef);
    Y2 = double(imageDis);
end

Y1 = double(Y1);
Y2 = double(Y2);
%%%%%%%%%%%%%%%%%%%%%%%%%% Download the image %%%%%%%%%%%%%%%%%%%%%%%%%
minDimension = min(rows,cols);
F = max(1,round(minDimension / 256)); 
aveKernel = fspecial('average',F); 
aveY1 = conv2(Y1, aveKernel,'same'); 
aveY2 = conv2(Y2, aveKernel,'same'); 
Y1 = aveY1(1:F:rows,1:F:cols); 
Y2 = aveY2(1:F:rows,1:F:cols); 
%%%%%%%%%%%%%%%%%%%%%%%%%% Calculate the gradient map%%%%%%%%%%%%%%%%%%%%%%%%%

dx = [1 0 -1; 1 0 -1;  1  0 -1]/3;
dy = [1 1 1; 0  0   0; -1 -1 -1]/3;%Prewitt operators

p=0.5;

IxY1 = conv2(Y1, dx, 'same');     
IyY1 = conv2(Y1, dy, 'same');    
gradientMap1 = abs(IxY1).^p +abs(IyY1).^p;

IxY2 = conv2(Y2, dx, 'same');     
IyY2 = conv2(Y2, dy, 'same');    
gradientMap2 = abs(IxY2).^p +abs(IyY2).^p;

%%%%%%%%%%%%%%%%%%%%%%%%%% Calculate the variance maps %%%%%%%%%%%%%%%%%%%%%%%%%
 [M N] = size(Y1);
% automatic downsampling
f = max(1,round(min(M,N)/256));
%downsampling by f use a simple low-pass filter 
if(f>1)
    lpf = ones(f,f);
    lpf = lpf/sum(lpf(:));
    Y1 = imfilter(Y1,lpf,'symmetric','same');
    Y2 = imfilter(Y2,lpf,'symmetric','same');

    Y1 = Y1(1:f:end,1:f:end);
    Y2 = Y2(1:f:end,1:f:end);
end
window = fspecial('gaussian', 11, 1.5);	
   L = 255; 			
   K(2) = 0.8;					 
  T2 = (K(2)*L)^2;
  T3=T2./2;
window = window/sum(sum(window));

mu1= filter2(window, Y1, 'same');
mu2= filter2(window, Y2, 'same');
mu1_sq = mu1.*mu1;
mu2_sq = mu2.*mu2;
mu1_mu2 = mu1.*mu2;
sigma1_sqMap1 = filter2(window, Y1.*Y1, 'same') - mu1_sq;
sigma2_sqMap2 = filter2(window, Y2.*Y2, 'same') - mu2_sq;
sigma12 = filter2(window, Y1.*Y2, 'same') - mu1_mu2;
CompareSimMatrix=(2*sigma1_sqMap1.^1/2.*sigma2_sqMap2.^1/2 + T2)./(sigma1_sqMap1 + sigma2_sqMap2 + T2);
StrucSimtMatrix=(sigma12 + T3)./(sigma1_sqMap1.^1/2.*sigma2_sqMap2.^1/2 + T3);


%%%%%%%%%%%%%%%%%%%%%%%%%% Calculate the GWL-SSIM %%%%%%%%%%%%%%%%%%%%%%%%%
 
T1=280;%fixed
gradientSimMatrix = (2*gradientMap1.*gradientMap2 + T1) ./(gradientMap1.^2 + gradientMap2.^2 + T1);
Gm = max(gradientMap1, gradientMap2);

SimMatrix =gradientSimMatrix .*  CompareSimMatrix.*StrucSimtMatrix.*Gm;

GWL_SSIM = sum(sum(SimMatrix)) ./ sum(sum(Gm));
quality_map=SimMatrix;
return;