【邮政编码识别】基于计算机视觉实现邮政编码识别含Matlab源码.zip

clc, clear, close all number_Image = 1; %用于处理的原图像编号. mkdir image; %新建子文件夹. addpath(genpath(pwd)); %将所有子文件夹添加到路径中. OriginalImage = imread(['image',num2str(number_Image),'.png']); figure subplot(211),imshow(OriginalImage),title('OriginalImage'); hsvImage = rgb2hsv(OriginalImage); %彩色图像转HSV以消除亮度和背景的影响. subplot(212),imshow(hsvImage),title('HsvImage'); hsv1 = hsvImage(:,:,1); hsv2 = hsvImage(:,:,2); hsv3 = hsvImage(:,:,3); figure %显示HSV三层图像 subplot(311),imshow(hsv1),title('Hsv1'); subplot(312),imshow(hsv2),title('Hsv2'); subplot(313),imshow(hsv3),title('Hsv3'); [m, n] = size(hsv1); d1 = zeros(m, n); d2 = zeros(m, n); for i = 1:m for j = 1:n if hsv3(i, j)<0.7 %hsv3层用来提取数字 d1(i, j) = 1; else d1(i, j) = 0; end if hsv2(i, j)>0.7 %hsv2层用来提取方格 d2(i, j) = 1; else d2(i, j) = 0; end end end figure % subplot(211),imshow(d2); bw1 = logical(d1);%double转logical型. thinImage = bwmorph(bw1, 'thin', Inf); %提取骨架. cleanImage = bwmorph(thinImage, 'clean', Inf); %去除孤立点. subplot(311),imshow(bw1),title('logicalImage of hsv2'); subplot(312),imshow(thinImage),title('thinImage'); subplot(313),imshow(cleanImage),title('cleanImage'); % SpurImage = bwmorph(g,'spur',3); %去毛刺. % figure % imshow(SpurImage); sum_rows = zeros(m,1); sum_columns = zeros(1,n); for i = 1:m for j = 1:n if d2(i,j)==1 sum_rows(i) = sum_rows(i) + 1; end end end boundary_rows_up = find(sum_rows>0, 1, 'first'); %上边界行标. boundary_rows_down = find(sum_rows>0, 1, 'last' ); %下边界行标. height = boundary_rows_down - boundary_rows_up +1; %方格高度. for i = 1:n for j = 1:m if d2(j,i)==1 sum_columns(i) = sum_columns(i) + 1; end end end boundary = zeros(m,n); %粗提左右边框 for i = boundary_rows_up:boundary_rows_down for j = 1:n if sum_columns(j)>m/2 boundary(i,j) = 1; end end end thinImage_boundary = bwmorph(boundary, 'thin', Inf); %提取左右边框骨架. n_boundary_c = zeros(1,n); %设立零矩阵存放图像每列的和. for i = 1:n for j = 1:m if thinImage_boundary(j,i)==1 n_boundary_c(i) = n_boundary_c(i) + 1; end end end boundary_columns = find(n_boundary_c>0); %提取方格左右边界. % len = size(boundary_columns); %必有12条边界 sum_boundary = 0; %求方格宽度. for i = 1:12 if mod(i,2)==0 sum_boundary = sum_boundary + boundary_columns(i) - boundary_columns(i-1); end end width = ceil(sum_boundary/6); %求平均宽度并向上取整. %提取每个方格并显示. figure for i = 1:12 %从原图中提取每个方格数据. if mod(i,2)==1 ExtractImage(:, :, (i+1)/2) = OriginalImage(boundary_rows_up:boundary_rows_down, boundary_columns(i):boundary_columns(i)+width); end end for i = 1:6 subplot(3,2,i); imshow(ExtractImage(:, :, i)); title(['number',num2str(i)]); BwImage_etc = ~im2bw(ExtractImage(:, :, i),0.7); %灰度图像二值化. ThinImage_etc = bwmorph(BwImage_etc, 'thin', Inf); %提取骨架. % SpurImage_etc = bwmorph(ThinImage_etc,'spur',4); %去毛刺. CleanImage_etc(:,:,i) = bwmorph(ThinImage_etc, 'clean'); %移除孤立点,若去毛刺则将ThinImage_etc改为SpurImage_etc. imwrite(CleanImage_etc(:,:,i), ['image/n',num2str(i),'.tif']); end %计算欧拉数，端点数，端点方向，并进行数字识别. figure % eul = zeros(1:6); % endp = zeros(1:6); for i = 1:6 subplot(3,2,i),imshow(['n',num2str(i),'.tif']),title(['n',num2str(i),'.tif']); eul(i) = bweuler(CleanImage_etc(:,:,i)); %计算欧拉数. x1 = CleanImage_etc(:,:,i); [endpointsg(:,:,i),endp(i)] = endpoints(x1,height,width+1); %求端点个数. end disp('euler = ');disp(eul); disp('endpoint = ');disp(endp); figure for i=1:6 subplot(3,2,i),imshow(endpointsg(:,:,i));title(['endpoint no.',num2str(i)]); %作端点图. end data = zeros(1,6); for i = 1:6 x1 = CleanImage_etc(:,:,i); x2 = endpointsg(:,:,i); data(i) = fcla(i,eul, endp, height, x1, x2); %提取图像特征并识别. end disp('data = ');disp(data);