无参考图像质量评价之基于多特征的增强图像质量评价

本节主要参考的文献是：

Leida, Shen, Jinjian, et al. No-Reference Quality Assessment of Enhanced Images[J]. 中国通信:英文版, 2016(9):121-130.

 主要思想

 本文主要是提取增强之后图像的多种特征来评价图像的质量分数，本文主要介绍相关的思想。

这篇文章主要提取五种主要的特征：非结构性特征--色彩连续性、色彩丰富度、对比度、光照度和熵。文中提出还用另外两个特征--锐 化度和自然特征，只不过这两种特征在其他论文中已经有介绍，就不介绍了。

 色彩连续性：对于彩色图像，先转换为HSI空间图像。对于某个像素，如果其色调（hue）跟周围像素的色调不一致，那么就认为这

色彩丰富度：高质量的图像，其色彩也更丰富。

 对比度：可以用图像直方图来表示，对比度越丰富，那么，直方图也就越均匀。感觉此处没有创新点

clc;clear all;load modelparameters.mat %Import variables  % mu_prisparam    - mean of multivariate Gaussian model% cov_prisparam   - covariance of multivariate Gaussian modelblocksizerow    = 96;%Height of the blocks in to which image is dividedblocksizecol    = 96;%Width of the blocks in to which image is dividedblockrowoverlap = 0;%Amount of vertical overlap between blocksblockcoloverlap = 0;%Amount of horizontal overlap between blockspath ='D:\0 Matlab\3 Matlab 图像处理\图像分级\niqe_release\niqe_release\test\';%Image pathN1=160;%Number of imagesfor L=1: N1    a = sprintf('%03d',L);%Similar to ‘0001’ as the prefix, the numbers increase in order    im=imread([path,'image',num2str(a),'.png']); %Read pictures continuously    quality(L)=computequality(im,blocksizerow,blocksizecol,blockrowoverlap,blockcoloverlap, ...        mu_prisparam,cov_prisparam);%Calculation of image qualityendMappedFlattened = mapminmax( quality, 0, 1); % The normalized.mkdir('D:\0 Matlab\3 Matlab 图像处理\图像分级\niqe_release\niqe_release\result\');% Create Image Save Pathpath1='D:\0 Matlab\3 Matlab 图像处理\图像分级\niqe_release\niqe_release\result\';%Image Save Pathfor i=1:length(MappedFlattened )    s=MappedFlattened(i);%Read normalized image quality    if s<=0.1        a = sprintf('%03d',i);%Similar to ‘0001’ as the prefix, the numbers increase in order        im=imread([path,'image',num2str(a),'.png']); %Read pictures continuously        imwrite(im,[path1,'1级_','image',num2str(a),'.png']);%save Picture    else if s<=0.30            a = sprintf('%03d',i);%Similar to ‘0001’ as the prefix, the numbers increase in order            im=imread([path,'image',num2str(a),'.png']); %Read pictures continuously            imwrite(im,[path1,'2级_','image',num2str(a),'.png']);%save Picture        else if s<=0.50                a = sprintf('%03d',i);%Similar to ‘0001’ as the prefix, the numbers increase in order                im=imread([path,'image',num2str(a),'.png']); %Read pictures continuously                imwrite(im,[path1,'3级_','image',num2str(a),'.png']);%save Picture            else if s<=0.7                    a = sprintf('%03d',i);%Similar to ‘0001’ as the prefix, the numbers increase in order                    im=imread([path,'image',num2str(a),'.png']); %Read pictures continuously                    imwrite(im,[path1,'4级_','image',num2str(a),'.png']);%save Picture                else                    a = sprintf('%03d',i);%Similar to ‘0001’ as the prefix, the numbers increase in order                    im=imread([path,'image',num2str(a),'.png']); %Read pictures continuously                    imwrite(im,[path1,'5级_','image',num2str(a),'.png']);%save Picture                end            end        end    endend

•