Volume: 17, Issue: 1, November 2013

========================================================================================================

Speed Sign Recognition using Independent Component Analysis

Authors: Ayman M Mansour, Jafar Abukhait, Imad Zyout

Abstract:

This paper proposes a system for speed limit signs recognition in the United States. The proposed system is based on Independent Component Analysis (ICA).  Our proposed system can be used in driver assistant system (DAS) or autonomous vehicles. The proposed system consists of four stages: 1) color segmentation to remove non-speed sign objects from the scene; 2) speed sign shape detection and recognition using geometric means; 3) Representation of speed sign images using Independent Component Analysis (ICA) ; 4) feature extraction and classification of speed sign images. The proposed system is invariant to scale, rotation, and partial occlusion. Independent Component Analysis (ICA) has been used in this paper to capture the inherent properties of the speed signs in order to be recognized. This was done by creating an independent component bank (or can be called as basis function bank) from a training set of speed signs. Feature vectors are generated from IC’s bank which is used for recognizing speed signs in the testing stage. Our experimental results show a significant recognition accuracy rate of rectangular speed signs.

(pp. 832-838)                    Download PDF

========================================================================================================