May I get the matlab code for Otsu's method of Thresholding (Gray Scale Image)?
3 views (last 30 days)
Show older comments
Answers (2)
Thomas
on 6 Dec 2012
The graythresh function uses Otsu's method, which chooses the threshold to minimize the intraclass variance of the black and white pixels.
another file exchange submission with the code is avialbel at
2 Comments
Thomas
on 6 Dec 2012
Edited: Thomas
on 6 Dec 2012
it is in the link in the file exchange submission.., asking for code is a very good way of getting your question ignored by regular contributors.. I am attaching the code form the file exchange submission this time..
this is the main otsu.m function
function [k]=otsu(a)
b=a;
[c,d]=size(b);
b=reshape(b,[],1);
[m,n]=size(b);
weightb=zeros(3,256);
weightf=zeros(3,256);
r=1;
l=1;
for T=0:255
[wb,wf,mb,mf,vrb,vrf]=cal(T,b,m);
weightb(1,r)=wb;
weightb(2,r)=mb;
weightb(3,r)=vrb;
r=r+1;
weightf(1,l)=wf;
weightf(2,l)=mf;
weightf(3,l)=vrf;
l=l+1;
end
%Within class variance
wcv=zeros(1,256);
for g=1:256
wcv(1,g)=((weightb(1,g)*weightb(3,g))+((weightf(1,g)*weightf(3,g))));
end
% min(wcv)
[threshold_value,val]=min(wcv);
tval=(val-1)/256;
% b=imresize(b,[c d])
a=im2bw(a,tval);
k= medfilt2(a,[25 25]);
end
This is the cal.m function thta calculate wb,wf,mb and so on
function [wb,wf,mb,mf,vrb,vrf]=cal(i,b,m)
% weight
wb=0;
wf=0;
mb=0;
mf=0;
b=double(b);
vb=0;vf=0;
vrb=0;
vrf=0;
for s=1:m
if(b(s,1)<(i))
wb=wb+1;
mb=mb+b(s,1);
else
wf=wf+1;
mf=mf+b(s,1);
end
end
%mean
if(wb==0)
mb=0;
mf=mf/wf;
elseif(wf==0)
wf=0;
mb=mb/wb;
else
mb=mb/wb;
mf=mf/wf;
end
%weight
wb=wb/m;
wf=wf/m;
%variance
for t=1:m
if(b(t,1)<(i))
vrb=vrb+((b(t,1)-mb)^2);
vb=vb+1;
else
vrf=vrf+((b(t,1)-mf)^2);
vf=vf+1;
end
end
if(vb==0)
vrb=0;
vrf=vrf/vf;
elseif(vf==0)
vrf=0;
vrb=vrb/vb;
else
vrb=vrb/vb;
vrf=vrf/vf;
end
See Also
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!