基于Matlab实现PCA与KPCA的SVM人脸识别系统+GUI操作界面的毕业设计（源码+数据集）.rar

%用KPCA进行人脸识别的Matlab代码 global h_axes1; global h_axes2; global edit2; h_f=figure('name','KPCA人脸识别系统','position',[300,200,600,400]); %clf reset;%clf 清除当前的图像的所有自图像，Reset重新设置图像的目标属性为默认值 set(h_f,'defaultuicontrolfontsize',12); set(h_f,'defaultuicontrolfontname','宋体'); h_axes1=axes('parent',h_f,'position',[0.2,0.28,0.25,0.56],'Unit','normalized','visible','on'); h_axes2=axes('parent',h_f,'position',[0.55,0.28,0.25,0.56],'Unit','normalized','visible','on'); figcolor=get(h_f,'color'); edit2=uicontrol(h_f,'style','text','position',[150,330,300,40],... 'backgroundcolor',figcolor);%动态变化提示 button_open=uicontrol(h_f,'style','push','string','选择照片'... ,'position',[250 40 100 50],'callback','GUIopen'); button_recg=uicontrol(h_f,'style','push','string','测试准确率',... 'position',[100 40 100 50],'callback','face'); button_match=uicontrol(h_f,'style','push','string','KPCA图像识别',... 'position',[400 40 100 50],'callback','GUIrecg'); set(edit2,'string','读取训练数据并训练....') drawnow allsamples=[];%所有训练图像 for i=1:40 for j=6:10 a=imread(strcat('Data\s',num2str(i),'\',num2str(j),'.pgm')); %a=imread(strcat('./Data/s',num2str(i),'/',num2str(j),'.jpg')); b=a(1:112*92); % b是行矢量 1×N，其中N＝10304，提取顺序是先列后行，即从上到下，从左到右 b=double(b); allsamples=[allsamples; b]; % allsamples 是一个200 * 10304 矩阵，allsamples 中每一行数据代表一张图片。 end end cov_size=size(allsamples,1); train_num=cov_size; patterns=allsamples; for i=1:cov_size for j=i:cov_size K(i,j) =(patterns(i,:)*patterns(j,:)'+1)^0.7; K(j,i) = K(i,j); %核函数是对称的 end end unit = ones(cov_size, cov_size)/cov_size; K_n = K - unit*K - K*unit + unit*K*unit; %改变核矩阵以实现数据在高维空间去均值 [evecs,evals] = eig(K_n); %求去均值后的核矩阵的特征值、特征向量 evals = real(diag(evals)); %取特征值实部 [l1,l2]=sort(-1*evals'); evals=l1*(-1); evecs=evecs(:,l2); % 取排序后特征值对应的特征向量 for i=1:cov_size, %因为K为满秩矩阵 evecs(:,i) = evecs(:,i)/(sqrt(evals(i))); %这样可以使特征空间中协方差矩阵C的特征向量归一化 end % 90%的贡献率 dsum = sum(evals); dsum_extract = 0; p = 0; while( dsum_extract/dsum < 0.9) p = p + 1; dsum_extract = sum(evals(1:p)); end max_ev=p; unit_train = ones(train_num,cov_size)/cov_size; for i=1:train_num, for j=1:cov_size, K_train(i,j) =(patterns(i,:)*patterns(j,:)'+1)^0.7; end end K_train_n = K_train - unit_train*K - K_train*unit + unit_train*K*unit; features = zeros(train_num, max_ev); features = K_train_n * evecs(:,1:max_ev); % 训练样本在特征空间的投影 allcoor = features; accu=0; % 测试过程 for i=1:40 for j=1:5 %读入40 x 5 副测试图像 a=imread(strcat('Data\s',num2str(i),'\',num2str(j),'.pgm')); %a=imread(strcat('./Data/s',num2str(i),'/',num2str(j),'.jpg')); b=a(1:10304); b=double(b); unit_test = ones(1,cov_size)/cov_size; K_test = zeros(1,cov_size); for j=1:cov_size, K_test(j) = (b*patterns(j,:)'+1)^0.7; end K_test_n = K_test - unit_test*K - K_test*unit + unit_test*K*unit; % 去均值 test_features = zeros(1, max_ev); test_features = K_test_n * evecs(:,1:max_ev); tcoor=test_features; %计算坐标，是1×p阶矩阵 for k=1:200 mdist(k)=norm(tcoor-allcoor(k,:)); %求2范数 end; [dist,index2]=sort(mdist); class=ceil( index2(1)/5 ); if class==i accu=accu+1; end; end; end; accuracy=accu/200;%输出识别率 set(edit2,'string','测试完成！') drawnow msgbox(['识别准确率：',num2str(accuracy*100),'%。']) global h_axes1 [filename,pathname]=uigetfile({'*.pgm';'*.jpg';'*.tif';'*.*'},'请选择一张用于识别的照片'); if filename==0 msgbox('请选择一张照片文件') else filepath=[pathname,filename]; axes(h_axes1); imshow(imread(filepath)); end disp(filename) global h_axes1 global h_axes2 global edit2 set(edit2,'string','读取待识别图片......') drawnow img=getimage(h_axes1);%获得之前选中的照片的信息 if isempty(img) msgbox('请先选择一张图片！') end % 测试图片 set(edit2,'string','识别中......') drawnow pause(5) %a=imread(strcat('Data\s',num2str(i),'\',num2str(j),'.pgm')); %a=imread(strcat('./Data/s',num2str(i),'/',num2str(j),'.jpg')); a=img; b=a(1:10304); b=double(b); unit_test = ones(1,cov_size)/cov_size; K_test = zeros(1,cov_size); for j=1:cov_size, K_test(j) = (b*patterns(j,:)'+1)^0.7; end K_test_n = K_test - unit_test*K - K_test*unit + unit_test*K*unit; % 去均值 test_features = zeros(1, max_ev); test_features = K_test_n * evecs(:,1:max_ev); size(test_features) tcoor=test_features; %计算坐标，是1×p阶矩阵 for k=1:200 mdist(k)=norm(tcoor-allcoor(k,:)); %求2范数 end; [dist,index2]=sort(mdist); class=ceil( index2(1)/5 ); set(edit2,'string','识别完成！') drawnow axes(h_axes2); imshow(imread(['Data\s',num2str(class),'\1.pgm'])); msgbox(['样本识别为第',num2str(class),'个人']) figure B=reshape(test_features',112,92); imshow(B,[]) %msgbox(['样本识别为第',num2str(class),'个人']) p;