1 / 16

Complex Networks for Representation and Characterization of Object

Complex Networks for Representation and Characterization of Object. For CS790g Project Bingdong Li 11/9/2009. Outline Of Methodology. Re-introduce Traditional Approaches Proposed Methods Issues Summary Questions and Comments. Re-introduce Traditional Approaches .

yvon
Download Presentation

Complex Networks for Representation and Characterization of Object

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. Complex Networks for Representation and Characterization of Object For CS790g Project Bingdong Li 11/9/2009

  2. Outline Of Methodology Re-introduce Traditional Approaches Proposed Methods Issues Summary Questions and Comments

  3. Re-introduce Traditional Approaches • External characteristics: • boundary and shape • chain codes, polygonal approximations, skeletons • Internal characteristics: • color and texture, • statistical approaches, structural approaches, spectral approaches • Both external and internal characteristics. Source: CS674 Image Processing Lecture

  4. Proposed Methods:Representation • Define the network • a regular lattice network in 2-D background, each node is connected to its nearest neighbors depending on the Euclidean distance • Each node is addressed by its normalized degree Source: CS674 Image Processing Lecture

  5. Proposed Methods:Representation • Two pixel A and B on the contour A <-->B <==> d(A, B) <= r and d(A, B) >= n • r is the shape control threshold • n is noise control threshold • d is the normalized distance • k is the average degree

  6. From Raw Image

  7. To Matrices

  8. Proposed Methods:Representation

  9. Proposed Methods:Characterization • Network similarity algorithm: structural similarities Mehler, Alexander(2008) 'STRUCTURAL SIMILARITIES OF COMPLEX NETWORKS: A COMPUTATIONAL MODEL BY EXAMPLE OF WIKI GRAPHS',Applied Artificial Intelligence,22:7,619 — 683

  10. Algorithm • Raw image • Segmentation • Build the complex network • Classifying the object using network similarity algorithm

  11. Expected Results • In most case, it will be a small world network

  12. Expected Results • A methods for object classification that • Leverage complex network technology • Represent the geometric information • Represent the spatial information • Invariant to rotation • Invariant to translation

  13. Issues • the photometric information was not represented • Thresholds

  14. Summary • In this project, we tried a new approach for representation and characterization of object • Firstly, traditional approaches and complex network related approaches are reviewed • Then, proposed a new methods, its definition of network, characterization, and algorithm to classify an object

  15. Questions and Comments

  16. Thanks

More Related