Dear experts
I want to detect the boundary of different subdomains in the attached image which are named as S1, S2, S3, S4. I have found an algorithm which is named as "burning algorithm" but i couldn't find any references which explain the method. Does anybody know about it? Is there any references or something which can help me?
Thank you in advance
Try looking at IIRC "percolation algorithms" ( why they call them that I do not know ) which is generally the class of algorithms used for detecting features like seams and cracks boundaries and so on. These provide a way to clean up the source imagery to get more of a line drawing as you have presented above. Once you have a line art version of the image, a map coloring algorithm is used to fill each region with a unique color. Then each color can be extracted as a region. Also there are vectorization algorithms which can take raster line art and convert them to a vector format. The vectorized data could be analyzed to construct a model of the closed regions.
Dear Mahar
I think the best algorithm for images has curvature is the Cuevelet Transform.
The literature is filled with the application of such transform.
Best wishes
Prof. S. El-Rabaie
- Badges
- Science topic
- Similar topics
- Computer Science
- Images