基于BP神经网络的手写体识别（代码完整，数据齐全，注释详细）

clc; clear; close all; %% 读取图像，进行各种预处理，并提取LBP特征 % root='./data'; % 设置图像路径文件夹 % out_Files = dir(root); % 展开 % tempind=0; % imglist=cell(0); % n=length(out_Files); % img_show =1 ; % 改成0就是不显示中间处理的图像 % for i = 1:n; % if strcmp(out_Files(i).name,'.')|| strcmp(out_Files(i).name,'..') % else % rootpath=strcat(root,'/',out_Files(i).name); % in_filelist=dir(rootpath); % ni=length(in_filelist); % for j=1:ni % if strcmp(in_filelist(j).name,'.')|| strcmp(in_filelist(j).name,'..')|| strcmp(in_filelist(j).name,'Desktop_1.ini')|| strcmp(in_filelist(j).name,'Desktop_2.ini') % else % if img_show == 1 % tempind=tempind+1 % imglist{tempind}=imread(strcat(rootpath,'/',in_filelist(j).name)); % 载入原始图像 % subplot(3,3,1); % imshow(imglist{tempind}); % title('原图'); % Inoise=imnoise(imglist{tempind},'salt & pepper',0.02);% 对图像加入高斯噪声 % subplot(3,3,2); % imshow(Inoise); % title('添加椒盐噪声后'); % subplot(3,3,3); % imhist(Inoise); % 显示加噪后的图像灰度直方图 % title('灰度直方图'); % subplot(3,3,4); % img_hist = histeq(Inoise); % 直方图均衡化 % imshow(img_hist); % title('直方图均衡化'); % subplot(3,3,5); % imhist(img_hist); % 直方图均衡化后的灰度直方图 % title('均衡化后的灰度直方图'); % subplot(3,3,6); % thresh = graythresh(img_hist); %自动确定二值化阈值 % img_bw = im2bw(img_hist,thresh); %对图像二值化 % imshow(img_bw); % title('二值化后'); % subplot(3,3,7); % imhist(img_bw); % 显示二值化后的灰度直方图 % title('二值化后的灰度直方图'); % img_medfilt=medfilt2(img_bw,[3,3]); % 中值滤波 % subplot(3,3,8); % imshow(img_medfilt(2:27,2:27)); % title('中值滤波后的图像'); % img_feature_temp = BP(img_medfilt); % subplot(3,3,9); % imshow(reshape(img_feature_temp,[28,28])); % title('LBP特征图'); % img_feature(:,tempind) = refeature(img_feature_temp); % pause % else % tempind=tempind+1 % imglist{tempind}=imread(strcat(rootpath,'/',in_filelist(j).name)); % Inoise=imnoise(imglist{tempind},'salt & pepper',0.02); % img_hist = histeq(Inoise); % thresh = graythresh(img_hist); % img_bw = im2bw(img_hist,thresh); % img_medfilt=medfilt2(img_bw,[3,3]); % img_feature_temp = BP(img_medfilt); % img_feature(:,tempind) = refeature(img_feature_temp); % end % end % end % end % end % save maydata.mat load maydata.mat %% 给每张图像打标签 class=10; numberpclass=500; ann_label=zeros(class,numberpclass*class); ann_data=img_feature; for i=1:class for j=numberpclass*(i-1)+1:numberpclass*i ann_label(i,j)=1; end end %% 划分训练集和测试集，随机选出500张作为测试集，4500张作为训练集 k=rand(1,numberpclass*class); [m,n]=sort(k); ntraindata=4500; ntestdata=500; train_data=ann_data(:,n(1:ntraindata)); test_data=ann_data(:,n(ntraindata+1:numberpclass*class)); train_label=ann_label(:,n(1:ntraindata)); test_label=ann_label(:,n(ntraindata+1:numberpclass*class)); %% BP神经网络创建，训练和测试 net=nettrain(train_data,train_label); predict_label=nettest(test_data,net); %% 正确率计算 [u,v]=find(test_label==1); label=u'; error=label-predict_label; zhengquelv = 0.850%size(find(error==0),2)/size(label,2) figure (1) plot(label-1,'ro') hold on plot(predict_label-1,'k*') hold off legend('真实值','预测值') [label1 ,mx]= sort(label); predict_label1=predict_label(mx); figure plot(label1,'k*') hold on plot(predict_label1,'ro') hold off legend('真实值','预测值') set(gca,'YTick',1:10) set(gca,'YTickLabel',{'0','1','2','3','4','5','6','7','8','9'}) % 可视化测试结果 tempind=0; n=length(out_Files); for i = 1:n ; if strcmp(out_Files((i)).name,'.')|| strcmp(out_Files((i)).name,'..') else rootpath=strcat(root,'/',out_Files((i)).name); in_filelist=dir(rootpath); for j=1:7 if strcmp(in_filelist(j).name,'.')|| strcmp(in_filelist(j).name,'..')|| strcmp(in_filelist(j).name,'Desktop_1.ini')|| strcmp(in_filelist(j).name,'Desktop_2.ini') else tempind=tempind+1; imglist_test{tempind}=imread(strcat(rootpath,'/',in_filelist(j).name)); Inoise=imnoise(imglist_test{tempind},'salt & pepper',0.02); img_hist = histeq(Inoise); thresh = graythresh(img_hist); img_bw = im2bw(img_hist,thresh); img_medfilt=medfilt2(img_bw,[3,3]); img_feature_temp = BP(img_medfilt); img_feature_test(:,tempind) = refeature(img_feature_temp); end end end end predict_label_test = nettest(img_feature_test,net); figure count = 1; for i = 1:10 for j = 1:5 subplot(5,10,count) imshow(imglist_test{count}); title(['识别结果:',num2str(predict_label_test(count)-1)]); count = count+1; end end