|Stochastic Systems Group|
Texture Based Image Segmentation
An information theoretic algorithm for segmenting an image using level set methods is presented in this talk. We focus on previous work done by Junmo Kim. After validating approximations made in his algorithm with theoretical results, we extend his work in two ways: a bias field estimator is developed to improve the scalar intensity-based segmentation method, and a vector-based method is developed to segment textured images. We use the framework of steerable pyramids to capture textures of an image used in this segmentation. Preliminary results on both artificial and natural images are presented.
Problems with this site should be emailed to email@example.com