基于图像的硬币检测与计数.zip

ph=imread('1y5j1j.jpg') image=ph; image1=rgb2hsv(image); I=image1(:,:,2); subplot(3,3,1);imshow(image1); bw=im2bw(I,0.4); subplot(3,3,2);imshow(I) subplot(3,3,3);imshow(bw) I1=imclose(bw,strel('disk',10)); I2=imopen(I1,strel('disk',10)); I3=bwareaopen(I2, 10000); subplot(3,3,4);imshow(I3); I4=imerode(I3,ones(1,1)); subplot(3,3,5);imshow(I4); [L,numa]=bwlabel(I4,8); image=ph; I=rgb2gray(image); figure;subplot(2,4,1);imshow(I); bw=im2bw(I,graythresh(I)); I1=bwareaopen(bw, 10000); I2=imclose(I1,strel('disk',5)); I3=imopen(I2,strel('disk',5)); subplot(2,4,2);imshow(I3) I4=medfilt2(I3,[5,5]); I5=bwareaopen(I4, 10000); subplot(2,4,3),imshow(I5); I6=edge(I5,'canny'); I7=imfill(I6 ,'holes'); I9=bwareaopen(~I7, 10000); I10=bwareaopen(~I9, 10000); subplot(2,4,4);imshow(I10) [L,T]=bwlabel(I10,8); if T==1 if numa==1; numa=1;c=0;i=0; disp(['一角的数量为']); disp(c); disp(['一元的数量为']); disp(i); disp(['五角的数量为']); disp(numa); else load templet; %加载数字特征 %A = imp; %得到待识别图片 Image=ph; I1= rgb2gray(Image); I9=im2bw(I1,graythresh(I1)); bw=imfill(~I9,'holes'); pictureCut = bwareaopen(bw, 1000); pictureRe = regionprops(pictureCut, 'area', 'boundingbox'); rects = cat(1, pictureRe.BoundingBox); %将面积对象的边界条件链接并保存到rects，顺序为[起始点x坐标, 起始点y坐标, 面积对象长度(x), 面积 rectangle('position', rects(1, :), 'EdgeColor', 'r'); %定位车牌区域，并用红色的框标记 pictureOut = imcrop(I1, rects(1, :)); %按照红线框切割车牌区域 p1=im2bw(pictureOut,graythresh(pictureOut)+0.05); SE = strel('disk',1) ; a=imerode(p1,SE); b=imdilate(p1,SE); p3=b-a; p=~p3; I4=imresize(p,[500 500]); figure(1),subplot(121),imshow(Image),title('待识别的图片：'); B=zeros(1,100); %创建1列100行的特征向量 [row,col] = size(I4); %得到待测样本的行列 cellRow = row/10; %将其降维10x10的特征图片 cellCol = col/10; count = 0; %每1/100个格子中为0的像素点个数 currentCell = 1; %当前计算为第1个1/100格子部分 for currentRow = 0:9 for currentCol = 0:9 for i = 1:cellRow %计算每1/100部分中为0的数量 for j = 1:cellCol if(I4(currentRow*cellRow+i,currentCol*cellCol+j)==0) count=count+1; end end end ratio = count/(cellRow*cellCol); %计算1/100部分中黑色像素的占比 B(1,currentCell) = ratio; %将每个占比统计在B特征向量中 currentCell = currentCell+1; %新的1/100部分开始计算 count = 0; %像素点计数置0 end end class = bayesBinary(B); %将该特征利用贝叶斯分类器分类,返回类的个数 subplot(122),imshow(['C:\Users\feng\Desktop\硬币\yangben\',num2str(class),'.jpg']),title('该数字被识别为：'); disp(['该数字被识别为：',pattern(class).name]); if class==1||class==4 c=0;i=1;numa=0; end if class==2||class==5 c=0;i=0;numa=1; end if class==3||class==6 c=1;i=0;numa=0; end disp(['一角的数量为']); disp(c); disp(['一元的数量为']); disp(i); disp(['五角的数量为']); disp(numa); end else I11=edge(I10,'sobel'); subplot(2,4,5);imshow(I11) stepr = 1; stepa = 0.1; minr =3; maxr =300; percent= 0.8; circlefind=findcircle(I11,minr,maxr,stepr,stepa,percent); [w,h]=size(circlefind); for g=1:w-1 for q=g+1:w if abs(circlefind(g,1)-circlefind(q,1))<5&&abs(circlefind(g,2)-circlefind(q,2))<5 circlefind(q,1)=0; circlefind(q,2)=0; circlefind(q,3)=0; end end end circlefind(all(circlefind==0,2),:)=[]; [j,k]=size(circlefind); i=0; [l,num]=bwlabel(I10,8); [my_max] = max(circlefind(:,3)); [my_min] = min(circlefind(:,3)); c=0; c1=c; i1=i; numa1=numa; if my_max-my_min>15 for n=1:j if my_max-circlefind(n,3)<10&&circlefind(n,3)>30 i1=i1+1;i=i+1; end end c1=T-numa1-i1; c=T-numa-i; disp(['一角的数量为']); disp(c1); disp(['一元的数量为']); disp(i1); disp(['五角的数量为']); disp(numa1); else for n=1:j if my_max-circlefind(n,3)<10&&circlefind(n,3)>30 i=i+1;i1=i1+1; end end c1=T-numa1; c=T-numa-i; disp(['一角的数量为']); disp(c1); disp(['一元的数量为']); disp(0); disp(['五角的数量为']); disp(numa1); end if c==0&&numa==0 %%c=0说明非五角大小相同 %利用贝叶斯分类器对手写图片识别 load templet; %加载数字特征 %A = imp; %得到待识别图片 Image=ph; I1= rgb2gray(Image); I9=im2bw(I1,graythresh(I1)); bw=imfill(~I9,'holes'); pictureCut = bwareaopen(bw, 1000); pictureRe = regionprops(pictureCut, 'area', 'boundingbox'); % areas = [pictureRe.Area]; %将面积对象保存到areas里 rects = cat(1, pictureRe.BoundingBox); %将面积对象的边界条件链接并保存到rects，顺序为[起始点x坐标, 起始点y坐标, 面积对象长度(x), 面积 rectangle('position', rects(1, :), 'EdgeColor', 'r'); %定位车牌区域，并用红色的框标记 pictureOut = imcrop(I1, rects(1, :)); %按照红线框切割车牌区域 p1=im2bw(pictureOut,graythresh(pictureOut)+0.05); SE = strel('disk',1) ; a=imerode(p1,SE); b=imdilate(p1,SE); p3=b-a; p=~p3; I4=imresize(p,[500 500]); figure(1),subplot(121),imshow(Image),title('待识别的图片：'); B=zeros(1,100); %创建1列100行的特征向量 [row,col] = size(I4); %得到待测样本的行列 cellRow = row/10; %将其降维10x10的特征图片 cellCol = col/10; count = 0; %每1/100个格子中为0的像素点个数 currentCell = 1; %当前计算为第1个1/100格子部分 for currentRow = 0:9 for currentCol = 0:9 for i = 1:cellRow %计算每1/100部分中为0的数量 for j = 1:cellCol if(I4(currentRow*cellRow+i,currentCol*cellCol+j)==0) count=count+1; end end end ratio = count/(cellRow*cellCol); %计算1/100部分中黑色像素的占比 B(1,currentCell) = ratio; %将每个占比统计在B特征向量中 currentCell = currentCell+1; %新的1/100部分开始计算 count = 0; %像素点计数置0 end end class = bayesBinary(B); %将该特征利用贝叶斯分类器分类,返回类的个数 subplot(122),imshow(['C:\Users\feng\Desktop\硬币\yangben\',num2str(class),'.jpg']),title('该数字被识别为：'); %disp(['该数字被识别为：',pattern(class).name]); if class==1||class==4 c=0;i=T;numa=0; end if class==2||class==5 c=0;i=0;numa=T; end if class==3||class==6 c=T;i=0;numa=0; end disp(['一角的数量为']); disp(c); disp(['一元的数量为']); disp(i); disp(['五角的数量为']); disp(numa); end end