I want to separate the clouds form the sky using the automatic graph cut method. Can anyone please help me. I read a paper which says we need to identify the hard constrained seeds first but I am unable to understand that. I am giving a sample image here.
Please try the threshold selection method first.
http://cn.mathworks.com/matlabcentral/fileexchange/56371-zhenzhou-threshold-selection
Also, the following references tell how to compute the seeds.
Zhenzhou Wang, “A new approach for segmentation and quantification of cells or nanoparticles” IEEE T IND INFORM,12(3):962-971, 2016.
Matlab codes are available at:
http://www.mathworks.com/matlabcentral/fileexchange/56365-generalized-framework-cell-segmentation
Dear Pradip,
Also i suggest to you links and attached files in topics.
-Automatic Graph Cut method - MATLAB Answers - MATLAB Central
https://fr.mathworks.com/.../325998-automatic-graph-cut-method
- Adaptive graph cut based cloud detection in wireless sensor networks
dl.acm.org/citation.cfm?id=2930111&CFID=882109544...
- CiteSeerX — Cloud Detection of RGB Color Aerial Photographs by ...
citeseerx.ist.psu.edu/viewdoc/citations;jsessionid...?doi=10.1.1...
-CiteSeerX — Cloud Detection of RGB Color Aerial Photographs by ...
citeseerx.ist.psu.edu/viewdoc/citations;jsessionid...?doi=10.1.1...
Best regards
Why do you expressly ask for graph-cut methods? If your goal is to detect and segment cloud as automatically and as reliably as you can, why restrict yourself to a single technique?
I suggest you look at the literature and chose among all techniques...
Article Cloud Detection and Classification with the Use of Whole-Sky...
Hello All. Thanks a lot for taking out your time to reply to my question. Although I am replying very late, I read all the answers on time and tested them.
@Zhenzhou Wang: Otsu algorithm does not perform well in all conditions. There are cases when half of the sky is dark and half is bright. In that case it marks some sky pixels as 1 and others as zero. So, I think for defining hard constraints otsu is not a good algorithm.
@Mohamed-Mourad Lafifi: Thanks a lot for the papers. Grow cut method works faster then Graph cut. Also, for the paper "Ground-Based Cloud Detection Using Automatic Graph Cut.pdf" I think the results are tailored for specific pictures. It does not give the desired results in all conditions. I wasted weeks in trying to implement their method, when I finally did, the method had flaws. They should really mention it in the paper.v Sadly they did not.
@Bruno Martin: Thanks for reply. This paper, along with Heinle paper I have read several times. The problem with the method is that it does not work well for all conditions. Using dual methods I think the problem of detecting thin and thick clouds can be over come. I have devised a new method for this and it is giving good results. Would share the report once its ready.
Hi Pradip Gupta. I need to help. I have a similar project . I want to calculate fraction of cloud in digital images. My hotmail is [email protected]. I need to talk with you. Thanks
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Algorithms