1

Please I'm working on Image Segmentation using Fuzzy C-means and K-means and C# Programming Language, But I will Like to know How I can Use C# to automatically determine the possible and best number of segment the image can be segmented into. Thanks

Al-Mustapha
  • 29
  • 4
  • Welcome to StackOverflow. Your question may be closed (disallowing answers) because it is not specific enough. Be as specific as possible. Explain what you have tried already. "How to use C# to.." is much too vague. For image processing questions, please ALWAYS post at least one sample image. http://stackoverflow.com/help/on-topic – Rethunk Nov 03 '13 at 02:30

1 Answers1

0

Determining the best number of image segments is a very difficult problem (perhaps even impossible without being very specific about what you mean by best). For example, if you handed an image to several people and asked them to manually segment it you would likely get many different results.

There are many different image segmentation algorithms. If you don't want to specify the number of clusters/segments as input, Fuzzy C-means or K-means are going to be poor choices.

Instead, I would suggest using a density based clustering algorithm. These algorithms don't require you to specify a number of clusters. Mean-shift clustering is one of these algorithms. I would suggest reading up on it.

See these posts:
Image Segmentation using Mean Shift explained
Python Clustering Algorithms

For more information about mean-shift: http://en.wikipedia.org/wiki/Mean-shift

Community
  • 1
  • 1
mattnedrich
  • 7,577
  • 9
  • 39
  • 45