二维LDA的matlab源代码

close all; clc; clear all; %load face database path=strcat('H:\深度数据\200305\') ; %%图像的路径是D盘得ORL，读取第i个人的第j个图像 nclass = 5; %数据库中有5个对象 nsmpaleeachclass = 16; %每个对象有1个人,分为16块 %图像集 for i=1:nclass for j = 1:nsampleeachclass X(:,(i-1)*nsampleeachclass+j) =facedatabase(:,(i-1)*(nsmpaleeachclass)+j*2-1); end end %测试样本 for i=1:nclass for j = 1:neachtest Xtest(:,(i-1)*neachtest+j) = facedatabase(:,(i-1)*(nsmpaleeachclass)+j*2); end end %训练样本矩阵转换为heigh*width的矩阵 for i=1:neachtrain*nclass trainSample{i}=reshape(Xtrain(:,i), height, width); %B = reshape(A,m,n) returns the m-by-n matrix B whose elements are taken column-wise from A. An error results if A does not have m*n elements end %测试样本矩阵转换为heigh*width的矩阵 for i=1:neachtest*nclass testSample{i}=reshape(Xtest(:,i), height, width); end [vec, val] = tdfda(trainSample, nclass); %计算矩阵的离散矩阵 for d=2:2:20 %对训练样本进行降维 for i=1:neachtrain*nclass %训练样本的总数 newTrainSample{i} = trainSample{i}*vec(:,1:d); newTestSample{i} = testSample{i}*vec(:,1:d); newXtrain(i,:) = reshape(newTrainSample{i}, 1, height*d); newXtest(i,:) = reshape(newTestSample{i}, 1, height*d); end %识别测试样本 %re_im{d/5} = uint8(newTrainSample{1}*vec(:,1:d)'); tic % toc Measure performance using stopwatch timer classification=classif(newXtrain', newXtest'); suc(d/2) = success(classification, neachtrain, neachtest); t(d/2)=toc; clear newTrainSample newTestSample newXtrain newXtest; end t disp('Recognition rate:'); suc box axis([2 20 0.65 1.00]); set(gca,'Xtick',[2:2:20],'ytick',[0.7:0.05:1.00]); ylabel('识别率'); xlabel('特征数'); hold on; plot(2:2:20,suc,'-^')