How do I call BLOCKPROC with more than one input image?
Afficher commentaires plus anciens
i have two images,and have a function which requires 3x3 sliding window data from both image A and image B. Is there a way to get it done with blockpro function. Any hint would be useful.
Réponses (2)
Image Analyst
le 8 Oct 2013
0 votes
See demos attached below.
4 commentaires
Sukuchha
le 9 Oct 2013
Image Analyst
le 10 Oct 2013
Your explanation is not clear. Why not just do blockproc on A and then do it on B? That would satisfy "a function which requires 3x3 sliding window data from both image A and image B". If you want something different, then say what is to be done at each block location to A and B.
Perhaps you can use nlfilter().
Sukuchha
le 15 Oct 2013
Image Analyst
le 15 Oct 2013
Sounds sort of complicated. What type of image are A and B? If they're color then you have 6 color channels on the cat'ed array, if they're grayscale, then you have 2. Usually image processing works with 1 or 3 color channels, not 2 or 6, so I'd be careful that it's doing what you expect. Why would you not do it like I said where you do it on A and then do it on B and then combine the results? That seems much more straightforward and less risky.
Ashish Uthama
le 10 Oct 2013
Modifié(e) : Ashish Uthama
le 10 Oct 2013
Have a look at the Definitions section of the blockproc doc. Consider using the 'location' and 'blockSize' fields to index into the second variable.
>> im1 = ones(10);
>> im2 = ones(10)*2;
>> type myfun.m
function o = myfun(bs, im2)
im1Block = bs.data;
startInd = bs.location;
endInd = bs.location+bs.blockSize-1;
im2Block = im2(startInd(1):endInd(1), startInd(2):endInd(2));
o = im1Block+im2Block;
end
% This is the tricky part. Here, all of im2 gets 'snapshot'ed inthe function handle, so its available inside myfun when called via funcHandle inside blockproc (!).
>> funcHandle = @(bs)myfun(bs,im2);
>> blockproc(im1,[5 5],funcHandle)
ans =
3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3
>>
1 commentaire
Sukuchha
le 15 Oct 2013
Catégories
En savoir plus sur Neighborhood and Block Processing dans Centre d'aide et File Exchange
Produits
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!