sobel 边缘检测算法

我的（但是，还需要做细化操作） W_H = [ 1 2 1; % the model in the horizon direction 0 0 0; -1 -2 -1]; W_V = [1 0 -1; % the model in the vetical direction 2 0 -2; 1 0 -1]; T = 0.165; % the threshold in the 2-value I = imread('f:\chinalake.bmp','bmp'); % read the image %I = imread('f:\rice.tif','tif'); [height,width] = size(I); I1 = double(I); I2 = zeros(height+2,width+2); % put the image's data into a bigger array to void the edge I2(2:height+1,2:width+1) = I1; for i=2:height+1 % move the window and calculate the grads for j=2:width+1 sum1 = 0; % the cumulus sum2 = 0; for m=-1:1 for n=-1:1 sum1 = sum1 + W_H(m + 2,n + 2) * I2(i + m,j + n); end end for m=-1:1 for n=-1:1 sum2 = sum2 + W_V(m + 2,n + 2) * I2(i + m,j + n); end end grey = abs(sum1) + abs(sum2); I1(i-1,j-1) = grey; end end big = max(max(I1)); % 归一化 small = min(min(I1)); for i=1:height for j=1:width I1(i,j) = (I1(i,j)-small)/(big - small); % 归一化 if(I1(i,j) > T) I1(i,j) = 1; % 二值化 else I1(i,j) = 0; end end end imshow(I1);