Vision Based Pedestrian Detection for Driver Assisted System

Abstract

This paper presents the research proposal of a Vision based pedestrian detection for driver assisted system. For this project, it was proposed to study vision based pedestrian detection by means of computational shape analysis and recognition where the 2D images are captured by using a 2D still camera. The computational analysis involves shape pre-processor, shape transformation and shape classification where each stage has specific computational techniques to be used. Finally, the pedestrian detection system will be tested and the results / findings should be able to reveal the possibility of detecting pedestrian using still camera as opposed to other techniques whereby the experiments involved the use of stereo vision, infra-red camera and microwave radar.