Otsu演算法進行映像二值化實現程式

Otsu演算法對映像進行二值化，核心就是計算背景和前景的最大類間方差，把類間方差最大的像素值作為全域二值化的閾值

二值化實現程式

%matlab程式<br />G = imread('IMG_0019.JPG');<br />I = rgb2gray(G);</p><p>[m,n] = size(I);<br />Hist = zeros(255);%長條圖<br />dHist = zeros(255);<br />variance = zeros(255);%方差<br />PXD = 0;</p><p>for i = 1:m<br /> for j = 1:n<br /> Hist(uint8(I(i,j))) = Hist(uint8(I(i,j))) + 1;<br /> end<br />end</p><p>for i = 1:255<br /> dHist(i) = Hist(i)/(m*n);<br />end</p><p>for PXD = 1:255<br /> w0 = 0;<br /> w1 = 0;<br /> g0 = 0;<br /> g1 = 0;<br /> for i = 1:PXD<br /> g0 = g0 + i*dHist(i);<br /> w0 = w0 + dHist(i);<br /> end<br /> for i = PXD+1 : 255<br /> g1 = g1 + i*dHist(i);<br /> w1 = w1 + dHist(i);<br /> end<br /> variance(PXD) = w0*w1*(g0 - g1)*(g0 - g1);<br />end</p><p>PXD = 1;<br />for i = 1:255<br /> if variance(PXD) < variance(i)<br /> PXD = i;<br /> end<br />end</p><p>for i = 1:m<br /> for j = 1:n<br /> if I(i,j) > PXD<br /> I(i,j) = 255;<br /> else<br /> I(i,j) = 0;<br /> end<br /> end<br />end<br />imagBW = I;<br />