基于PCA的ORL数据库人脸识别系统附matlab代码.zip

%% Slightly Modified LOOCV Test: Performed after changing orignal mean %% operation in PCA_NEW function with [mode, median] and comparision with %% mean-PCA (Orignal) the function shows time and recognition percentage function []=Comparision(); clc close all clear all %% Loading LOOCV Results if (exist('PCA_mean_LOOCV_Results\Average_LOOCV.mat','file') && exist('PCA_mean_LOOCV_Results\R.mat','file') && exist('PCA_median_LOOCV_Results\Average_LOOCV.mat','file') && exist('PCA_median_LOOCV_Results\R.mat','file')) load('PCA_mean_LOOCV_Results\Average_LOOCV.mat'); load('PCA_mean_LOOCV_Results\R.mat'); Mean_LOOCV_Results=Average_LOOCV; R_Mean=R; load('PCA_median_LOOCV_Results\Average_LOOCV.mat'); load('PCA_median_LOOCV_Results\R.mat'); Median_LOOCV_Results=Average_LOOCV; R_Median=R; load('PCA_mode_LOOCV_Results\Average_LOOCV.mat'); load('PCA_mode_LOOCV_Results\R.mat'); Mode_LOOCV_Results=Average_LOOCV; R_Mode=R; %% Graphing LOOCV Recognition Results figure('Name','Recognition Percentage vs Number of PCA Dimensions','NumberTitle','off') hold on [val1,ind1]=max(Mean_LOOCV_Results); plot(1:dimensions,Mean_LOOCV_Results,'LineWidth',3); plot(ind1,Mean_LOOCV_Results(ind1),'o','LineWidth',2,... 'MarkerEdgeColor','r',... 'MarkerFaceColor','g',... 'MarkerSize',8); [val2,ind2]=max(Median_LOOCV_Results); plot(1:dimensions,Median_LOOCV_Results,'--r','LineWidth',3); plot(ind2,Median_LOOCV_Results(ind2),'o','LineWidth',2,... 'MarkerEdgeColor','y',... 'MarkerFaceColor','k',... 'MarkerSize',8); [val3,ind3]=max(Mode_LOOCV_Results); plot(1:dimensions,Mode_LOOCV_Results,'g','LineWidth',3); plot(ind3,Mode_LOOCV_Results(ind3),'o','LineWidth',2,... 'MarkerEdgeColor','g',... 'MarkerFaceColor','m',... 'MarkerSize',8); grid minor title('Recognition Percentage vs Number of PCA Dimensions'); xlabel('Number of PCA Dimensions'); ylabel('Recognition Percentage'); legend('Recognition Percentage Using Orignal PCA',strcat('Maximum Recogntion :',num2str(val1),'% Using Orignal PCA'),'Recognition Percentage Using Improved(Median) PCA',strcat('Maximum Recogntion :',num2str(val2),'% Using Improved(Median) PCA'),'Recognition Percentage Using Mode PCA',strcat('Maximum Recogntion :',num2str(val3),'% Using Mode PCA'),'Location','Best') %% Graphing LOOCV Time/Dimensions Results figure('Name','Recognition Time vs Number of PCA Dimensions','NumberTitle','off') hold on for i = 1:dimensions mean_time1(i)=mean(R_Mean{1,i}(:,2)); end mean_time1=(mean_time1/ns)*1000; % v=mean(mean_time1); % ul=v+std(mean_time1); % ll=v-std(mean_time1); plot(mean_time1,'m','LineWidth',2) % plot(1:dimensions,repmat(ul,1,dimensions),'--k','LineWidth',2); % plot(1:dimensions,repmat(v,1,dimensions),'--r','LineWidth',2); % plot(1:dimensions,repmat(ll,1,dimensions),'--k','LineWidth',2); plot(ind1,mean_time1(ind1),'o','LineWidth',2,... 'MarkerEdgeColor','r',... 'MarkerFaceColor','g',... 'MarkerSize',8) for i = 1:dimensions mean_time2(i)=mean(R_Median{1,i}(:,2)); end mean_time2=(mean_time2/ns)*1000; % v=mean(mean_time2); % ul=v+std(mean_time2); % ll=v-std(mean_time2); plot(mean_time2,'LineWidth',2) % plot(1:dimensions,repmat(ul,1,dimensions),'--r','LineWidth',2); % plot(1:dimensions,repmat(v,1,dimensions),'--k','LineWidth',2); % plot(1:dimensions,repmat(ll,1,dimensions),'--r','LineWidth',2); plot(ind2,mean_time2(ind2),'o','LineWidth',2,... 'MarkerEdgeColor','r',... 'MarkerFaceColor','k',... 'MarkerSize',8) for i = 1:dimensions mean_time3(i)=mean(R_Mode{1,i}(:,2)); end mean_time3=(mean_time3/ns)*1000; % v=mean(mean_time2); % ul=v+std(mean_time2); % ll=v-std(mean_time2); plot(mean_time3,'g','LineWidth',2) % plot(1:dimensions,repmat(ul,1,dimensions),'--r','LineWidth',2); % plot(1:dimensions,repmat(v,1,dimensions),'--k','LineWidth',2); % plot(1:dimensions,repmat(ll,1,dimensions),'--r','LineWidth',2); plot(ind3,mean_time3(ind3),'o','LineWidth',2,... 'MarkerEdgeColor','m',... 'MarkerFaceColor','k',... 'MarkerSize',8) grid minor title('Mean Recognition Time per image vs Number of PCA Dimensions'); xlabel('Number of PCA Dimensions'); ylabel('Mean Recognition Time per image in (miliseconds)'); legend('Mean Recognition Time of Orignal PCA',strcat('Mean Maximum Recognition time :',num2str(mean_time1(ind1)),' milisec'),'Mean Recognition Time of Improved(Median) PCA',strcat('Mean Maximum Recognition time :',num2str(mean_time2(ind2)),' milisec'),'Mean Recognition Time of Mode PCA',strcat('Mean Maximum Recognition time :',num2str(mean_time3(ind3)),' milisec')); %% Results Summary fprintf('Using Orignal PCA %.2f %% Maximum Recognition is achieved at %d number of dimensions for mean recogntion time per image %f miliseconds\n\n',val1,ind1,mean_time1(ind1)); fprintf('Using Improved PCA %.2f %% Maximum Recognition is achieved at %d number of dimensions for mean recogntion time per image %f miliseconds\n\n',val2,ind2,mean_time2(ind2)); fprintf('Using Mode PCA %.2f %% Maximum Recognition is achieved at %d number of dimensions for mean recogntion time per image %f miliseconds\n\n',val3,ind3,mean_time3(ind3)); else fprintf('Perform LOOCV Test first\n'); end