K-L变换实验.docVIP

K-L变换 clc; clear all; A1=imread(C:\Users\Administrator\Desktop\模式识别用图\s8_1.bmp) A2=imread(C:\Users\Administrator\Desktop\模式识别用图\s8_2.bmp) A3=imread(C:\Users\Administrator\Desktop\模式识别用图\s8_3.bmp) B1=imread(C:\Users\Administrator\Desktop\模式识别用图\s10_1.bmp) B2=imread(C:\Users\Administrator\Desktop\模式识别用图\s10_2.bmp) B3=imread(C:\Users\Administrator\Desktop\模式识别用图\s10_3.bmp) C1=imread(C:\Users\Administrator\Desktop\模式识别用图\s30_1.bmp) C2=imread(C:\Users\Administrator\Desktop\模式识别用图\s30_2.bmp) C3=imread(C:\Users\Administrator\Desktop\模式识别用图\s30_3.bmp) S1=reshape(A1,112*92,1); S2=reshape(A2,112*92,1); S3=reshape(A3,112*92,1); S4=reshape(B1,112*92,1); S5=reshape(B2,112*92,1); S6=reshape(B3,112*92,1); S7=reshape(C1,112*92,1); S8=reshape(C2,112*92,1); S9=reshape(C3,112*92,1); S=[S1,S2,S3,S4,S5,S6,S7,S8,S9]; S=double(S); R=S*S; [V,d]=eig(R) u=S*V/sqrt(D); V=fliplr(V); D=fliplr(D); U=S*V; U=orth(U);E=sum(D); G=sum(E,2); N=3 for i=1:N E=sum(E(1:i),2); ratio=E/G; if ratio0.99 break; end end A= U(:,1:i); y= A*S; B=zeros(6,N); for i=1:N for j=4:9 Test=S(:,(i-1)*9+j)-J; H=A*Test; %求得其在特征脸空间的坐标 for k=1:N C=H-y(:,k); K(:,k)=C*C; end [h,I]=min(K); %最近邻法 if I==i B(j-3,i)=1; else B(j-3,i)=0; end %图像是否识别，识别出来为1，未识别出来为0 end end D=sum(B); D=sum(D,2); Rate=D/(6*N) %识别正确率 五、实验结果及结论： 1、d = 1.0e+09 * 0.0016 0 0 0 0 0 0 0 0 0 0.0022 0 0 0 0 0 0 0 0 0 0.0031 0 0 0 0 0 0 0 0 0 0.0039 0 0 0 0 0 0 0 0 0 0.0065 0 0 0 0 0 0 0 0 0 0.0109 0 0 0 0 0 0 0 0 0