Indexed by:会议论文

Date of Publication:2009-12-19

Included Journals:EI、CPCI-S、Scopus

Page Number:1172-+

Abstract:Efficiently and accurately detecting pedestrian plays a very important role in many computer vision applications such as Intelligent Transportation System and Safety Driving Assistant. This paper puts forwards a two-stage pedestrian detection method based on machine vision. Firstly, the expanded Haar-like characteristic is selected and calculated using integral map and the pedestrian detection cascaded classifiers with high accuracy are trained by Adaboost. After segmenting the candidate pedestrian areas from the image, a confirmation step is needed to judge whether those areas are pedestrian or not. Through analyzing the sample images, we can know that the gray image of pedestrian has some texture and gray symmetry features. In addition, the continuous edges of pedestrian make the extracted edges have certain boundary moments and gradient direction characters. Based on these features, each sample image is expressed by a multi-dimension characteristic vector. The final pedestrian classifier is obtained using support vector machines (SVM) training with the features abstracted above. The experiment results indicate that the algorithm could achieve effective recognition of vehicle proceeding pedestrians with different sizes, colors and shapes.