1 / 24

Image Segmentation based on multicue fusion

Image Segmentation based on multicue fusion. XiaoweiGeng 2007.09.24. Content. Introduction about image segmentation image segmentation priciples Main Process Related Results. Main Principles. Conditional Random Fields Model Mean Shift For color likelihood estimation

Download Presentation

Image Segmentation based on multicue fusion

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Image Segmentation based on multicue fusion XiaoweiGeng 2007.09.24

  2. Content • Introduction about image segmentation • image segmentation priciples • Main Process • Related Results

  3. Main Principles • Conditional Random Fields Model • Mean Shift For color likelihood estimation • Gabor Filters For texture likelihood estimation • Energy Optimized by Graph Cuts

  4. Conditional Random Fields • Definition (1) G=(V,E) is a graph (2) Y is indexed by the vertices of G . (3) the random variables obey the Markov property with respect to the graph, when conditioned on X : (X,Y) is a conditional random field. where means that w and v are neighbors in G.

  5. Conditional Random Fields • Characteristics where x is data information,y is label information,and y|s is the set of components of y associated with the vertices in subgraph S.

  6. Kernel Density Estimation (1) Kernel density estimator (2) two methods of H chosen in practice

  7. Mean Shift • A method to find modes of function

  8. Gabor Filters • Biological enlightenment • The Gabor Filters characteristics

  9. Gabor Filters

  10. Image Segmentation based on CRF • Construct Energy Function

  11. Energy Function • Our Energy Function As the similar presentation of MRF,we still use the classification to explain every term.

  12. Color Likelihood Term • Similar to Bayesian explanation,we define Here ,xi is the mean value of background or foreground sample

  13. Texture Likelihood Term • First , get the feature image by Gabor filters • second, use mean shift to filter the interesting regions to get the mean feature vector • Third ,use kernel density estimation to get the texture likelihood term

  14. Gabor Filter banks

  15. Gabor Filters

  16. Smooth Term • Smooth Term in order to make the results more smooth,we define the color smooth and texture smooth Terms as:

  17. Energy Optimize • Graph Cuts

  18. Some Results

  19. Some Results

  20. Some Results

  21. Some Results

  22. Some Results

  23. Some Results

More Related