基于matlab的植物病虫害图像识别.zip

function kkk = tiqubeijing clear clc I_rgb=imread('C:\Users\Administrator\Desktop\target172.png'); figure(1);subplot(341),imshow(I_rgb);title('原始图像'); %-------------------------------------------------------------------------- %去噪 filter=ones(5,5);%卷积核 filter=filter/sum(filter(:));%均值滤波核函数 denoised_r=conv2(I_rgb(:,:,1),filter,'same');%same表示输出大小与输入同，R信道滤波 denoised_g=conv2(I_rgb(:,:,2),filter,'same'); denoised_b=conv2(I_rgb(:,:,3),filter,'same'); denoised_rgb=cat(3, denoised_r, denoised_g, denoised_b);%3维合并矩阵 D_rgb=uint8(denoised_rgb);%变换数据类型到0~255 subplot(342),imshow(D_rgb);title('去噪后图像');%去噪后的结果 %-------------------------------------------------------------------------- %将彩色图像从RGB转化到lab彩色空间, 因为lab空间相互分量联系性比较小 利于分割 C =makecform('srgb2lab'); %设置转换格式 I_lab= applycform(D_rgb, C); %-------------------------------------------------------------------------- %进行K-means聚类将图像分割成2个区域 ab =double(I_lab(:,:,2:3)); %取出lab空间的a分量和b分量 nrows= size(ab,1); ncols= size(ab,2); ab =reshape(ab,nrows*ncols,2);%改变矩阵维数 nColors= 2; %分割的区域个数为2 [cluster_idx,cluster_center] =kmeans(ab,nColors,'distance','sqEuclidean','Replicates',2); %重复聚类2次 %Distance，距离测度。sqEuclidean，欧式距离。Replicates，整数。 pixel_labels= reshape(cluster_idx,nrows,ncols); %-------------------------------------------------------------------------- %显示分割后的各个区域 segmented_images= cell(1,3);%构造了1行3列的cell类型的矩阵 rgb_label= repmat(pixel_labels,[1 1 3]);%用矩阵pixel_labels复制[1 1 3] for k= 1:nColors color = I_rgb;%还原到RGB空间 color(rgb_label ~= k) = 0; segmented_images{k} = color; end subplot(343),imshow(segmented_images{1}),title('提取背景'); subplot(344),imshow(segmented_images{2}),title('提取叶片'); segmented_images{2}(segmented_images{2}~=0)=255; I0=im2bw(rgb2gray(segmented_images{2})); subplot(345),imshow(I0),title('二值图') %-------------------------------------------------------------------------- %标记最大连通区域 L = bwlabel(I0);%标记连通,这个函数是求它的局部属性 stats = regionprops(L); Ar = cat(1, stats.Area); ind = find(Ar ==max(Ar)); I1=zeros(size(I0)); I1(L == ind) = 1; I0(L == ind) = 0; subplot(346),imshow(I1,[]),title('连通面积最大的区域') %-------------------------------------------------------------------------- %-------------------------------------------------------------------------- %填充孔洞操作 I1=imfill(I1,'holes'); subplot(3,4,7), imshow(I1);title('填充孔洞操作'); %去叶柄 SE = strel('rectangle',[6 5]);%生成矩形结构元素 BW2 = imerode(I1,SE);%对图像进行腐蚀 subplot(348),imshow(BW2); title('腐蚀后的图像'); L = bwlabel(BW2);%标记连通 stats = regionprops(L); Ar = cat(1, stats.Area); ind = find(Ar ==max(Ar)); I1=zeros(size(I0)); I1(L == ind) = 1; I0(L == ind) = 0; subplot(349),imshow(I1,[]),title('连通面积最大的区域') BW2 = imdilate(I1,SE);%对图像进行膨胀 subplot(3,4,10), imshow(BW2);title('先腐蚀后膨胀的图像'); %-------------------------------------------------------------------------- %轮廓追踪 [i,j]=size(BW2); BW4=zeros(i,j); B=bwboundaries(BW2);%返回边界和标签矩阵 for k=1:length(B) boundary= B{k}; end [a,b]=size(boundary);%a是边界周长 for k=1:a BW4(boundary(k,1),boundary(k,2))=1; end subplot(3,4,11),imshow(BW4);title('边界追踪'); subplot(3,4,12),imshow(BW4);title('轮廓放大'); L1= bwlabel(BW2);%标记连通 start1 = regionprops(L1); kkk=BW4; %-------------------------------------------------------------------------- %标记最小包围盒 figure(2) imshow(BW4);title('局部属性'); hold on [rectx,recty,area,perimeter] = minboundrect(boundary(:,1),boundary(:,2),'a'); % 'a'是按面积算的最小矩形，如果按边长用'p'。 c1=sqrt((rectx(1)-rectx(2))*(rectx(1)-rectx(2))+(recty(1)-recty(2))*(recty(1)-recty(2)));%最小包围盒长度或者宽度 c2=sqrt((rectx(2)-rectx(3))*(rectx(2)-rectx(3))+(recty(2)-recty(3))*(recty(2)-recty(3)));%最小包围盒长度或者宽度 plot(recty,rectx,'-r'); %-------------------------------------------------------------------------- %标记凸包 x=boundary(:,1); y=boundary(:,2); k = convhull(x,y); plot(y(k),x(k), '-b'); hold on x1=y(k);%第一列横坐标赋值 y1=x(k);%第二列纵坐标赋值 n=length(x1);%点的个数 for i=1:(n-1) %计算各边长 p(1,i)=sqrt((x1(i)-x1(i+1))*(x1(i)-x1(i+1))+(y1(i)-y1(i+1))*(y1(i)-y1(i+1))); end totalare1=sum(p(1,:));%不规则图像的周长 N = length(x1); interv_x = zeros( 1, N - 1 ); mid_y = zeros( 1, N - 1); for i = 2 : length(x1) mid_y(i-1) = ( y1(i) + y1(i-1) ) / 2; interv_x(i - 1) = x1(i) - x1(i-1); end totalare= interv_x * mid_y';%凸包面积 %-------------------------------------------------------------------------- %标记质心 L = bwlabel(BW4);%标记连通 stars = regionprops(L); e=cat(1,stars.Centroid); plot(e(:,1),e(:,2),'w*'); for q=1:a t(q,1)=sqrt((boundary(q,2)-e(1,1))*(boundary(q,2)-e(1,1))+(boundary(q,1)-e(1,2))*(boundary(q,1)-e(1,2)));%质心到边界的距离 end p1=min(t);%质心到边界最小距离 p2=max(t);%质心到边界最大距离 sita=0:pi/20:2*pi; %plot(r*cos(sita),r*sin(sita)); %中心点在原点，半径为r的圆 plot(e(1,1)+p1*cos(sita),e(1,2)+p1*sin(sita),'c');%中心点在（x0,y0）半径为p1的圆 plot(e(1,1)+p2*cos(sita),e(1,2)+p2*sin(sita),'g');%中心点在（x0,y0）半径为p2的圆 %-------------------------------------------------------------------------- f1=start1.Area;%叶片面积 f2=a;%叶片周长 f3=max(c1,c2);%最小包围盒长度 f4=min(c1,c2);%最小包围盒宽度 f5=totalare;%凸包面积 f6=totalare1;%凸包周长 f7=p1;%内切圆半径 f8=p2;%外切圆半径 %f9=stars.Majoraxislength;%与区域具有相同标准的二阶中心距的椭圆长轴 %f10=stars.Minoraxislength;%与区域具有相同标准的二阶中心距的椭圆短轴 th1=f3/f4; th2=f1/f5; th3=f2/f6; th4=f7/f8; th5=4*pi*f1/(f6*f6); th6=f8/f7; th7=4*pi*f1/(f2*f2); th8=f1/(f3*f4); %th9=f9/f10; str1=['纵横中比=',num2str(th1)]; str2=['面积凹凸比=',num2str(th2)]; str3=['周长凹凸比=',num2str(th3)]; str4=['球形性=',num2str(th4)]; str5=['圆形性=',num2str(th5)]; str6=['偏心率=',num2str(th6)]; str7=['形状参数=',num2str(th7)]; str8=['矩形度=',num2str(th8)]; %str9=['等二阶矩椭圆长短轴比=',num2str(th9)]; %gtext(str1) %gtext(str2) %gtext(str3) %gtext(str4) %gtext(str5) %gtext(str6) %gtext(str7) %gtext(str8)