Loading...

Multiclass Visual Object Recognition Based On Cluttered Images

Moghimi Najafabadi, Mohammad | 2010

543 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 40713 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Kasaei, Shohreh
  7. Abstract:
  8. With the advancement of Machine Vision and Image Processing systems, the need for conceptual interpretation is raising. For this interpretation, one should detect objects available in the image and then tries to find the relations between the objects. For a good interpretation of the image, the machine vision system should learn patterns available in the nature. Object recognition systems are also used in other vision tasks and they can be used for content-based image retrieval, control and surveillance, or human action and gesture recognition for a better and easier human-computer interface. In an object recognition system, first some features should be extracted from the input image and then these features should be fed to an object recognition model to find its true class which it belongs to. In this thesis, different methods in feature extraction and recognition models are implemented and the results of the implementation are presented. Then, a method based on using inter-class relations is presented. This method can be used to automatically find inter-class relations and these relations are used to improve classification. Experimental results on Caltech-101 dataset show that, having similar accuracy, this method outperforms best object recognition systems in terms of speed and can be used in huge datasets. Another contribution of this research is proposing a method which helps to solve object shift in the input images. The proposed method extends a method called spatial pyramid match kernel. It improves the accuracy of the spatial pyramid matching kernel by approximately 35% in the case of shifted objects. In this thesis, first object recognition methods are discussed, then researches done, implementation issues, and experimental results are presented.

  9. Keywords:
  10. Machine Vision ; Object Recognition ; Multiclass Classification ; Local Invariant Feature

 Digital Object List

 Bookmark

No TOC