ICIP 2006, Atlanta, GA

Slide Show

Atlanta Conv. & Vis. Bureau


Technical Program

Paper Detail

Session:Edge Detection and Image Segmentation
Time:Monday, October 9, 14:20 - 17:00
Presentation: Poster
Authors: Farhang Sahba; University of Waterloo 
 Hamid R. Tizhoosh; University of Waterloo 
 Magdy M. A. Salama; University of Waterloo 
Abstract: A new approach to object extraction and recognition based on reinforcement learning is presented. We use this novel idea as an effective way to optimally segment the image and increase recognition rate. This new approach is introduced to segment a specific object, which should be recognized. The proposed method controls the segmentation and the post-processing based on an intelligent scheme. The main purpose of this work is to demonstrate this potential that an ”object-dependent segmentation” via an intelligent technique outperforms the best thresholding algorithms with respect to final recognition rate. It is important to notice that we chose reinforcement learning, as an intelligent technique, since it does not need a huge training data and can be applied if no expert or a-priori knowledge is available. Promising results showed the effectiveness of proposed approach.