Analyzing a large orthophoto block by block

I am trying to apply an algorithm to an orthophoto (6560x5550x3). Since it has lot of details in it, i want to analyze it in parts. I've tried using "blockproc function" in MATLAB, but i am not able to get any block of original orthophoto.
If there is some other way apart from blockproc then please suggest.
Thanks

Answers (1)

I don't know why. Can you explain why? Who cares if your algorithm have lots of details - just call nlfilter(). Any reason why nlfilter() can't do the job?
I also have some blockproc demos for you that I can attach, if I hear from you a good reason why you require blockproc().

6 Comments

I am working on large size image which is basically an aerial view image of an area wherein i want to detect rooftops of buildings and find severl more parameters related to the identified rooftops.
Since there is too much detail in it, i want to apply my algorithm to certain block on the image and lie this gradually all over the image. Image is too large and i cannot attach here.
If you could provide tutorial link to blockproc or any example illustrating , it would be very helpful.
Thanks
What is the size of your image? Also be aware that normally blockproc() processes the image in "jumps" and so you'll have reduced resolution in the output image. The demos are attached. But I'm still not convinced that blockproc is the correct approach. I don't see why a regular scanning window approach won't work.
Size of the image is 20Mb . dimension is 6560 x 5550.
Can you answer the last sentence in my comment above? And I have two new ones. What is the algorithm that you want to apply to each block? Did you adapt my any of my blockproc demos to you code?
Sir, The algorithm accounts for shadow, roof surface available (bright and not), roof features (i.e. chimneys or walls) and azimuthal angle of the buildings in the image. Basically, i am identifying only the rooftops of buildings.
I tried using blockproc_demos but it says out of memory everytime. I will try the normal scanning window approach.
Please tell me the new ones.
That does not seem like a reason to use blockproc(). But whatever...good luck.

Sign in to comment.

Asked:

on 17 Jun 2014

Commented:

on 21 Jun 2014

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!