Image Matching
- From: "Salha" <kuwaitisalha@xxxxxxxxx>
- Date: Sun, 26 Oct 2008 13:00:04 +0000 (UTC)
Hello There,
Am developing an application that is based on feature-based image matching.
1. I have two different images which I calculate for it corners by harries corner detector (PIP,PIP1).
2. Then I calculate the distance between these corners
3. I assign the distance matrix to a bipartite matching code to find the best match
The problem is that I will get best match of these points knowing that the 2 images are different!! I guess: I have to assess the distance values before passing it to the bipartite matching, maybe!
But I have no idea how to asses!
X1=double(img1);
X1=double(img2);
window_size = 5;
Size_PI = size(PIP,1);
Size_PI1 = size(PIP1,1);
for j=1:Size_PI
A = X1(PIP(j,1)-window_size:PIP(j,1)+window_size, PIP(j,2)-window_size:PIP(j,2)+window_size,:);
for i=1:Size_PI1
B = X2(PIP1(i,1)-window_size:PIP1(i,1)+window_size, PIP1(i,2)-window_size:PIP1(i,2)+window_size,:);
%distance btwn the points
dist(j,i) = sum(sum(sum(abs(A-B))));
end
end
%bipartite matching
best_dist= assignmentoptimal(dist);
*btw, this code will work perfectly with two semi-similar images!
.
- Prev by Date: Re: How can I "granulate" an image?
- Next by Date: 2D circular convolution matrix
- Previous by thread: help about writing a m file
- Next by thread: 2D circular convolution matrix
- Index(es):
Relevant Pages
|
