Research On Pedestrian Tracking Algorithm Based On Multiple Features And Particle Filter

Vehicle and pedestrian’s collision is the main type of road traffic accidents and pedestrian is always in weak position. Carrying out detection of pedestrian in critical condition and developing a timely warning system can effectively reduce the number of pedestrian casualties in road traffic, and will make great contributions on promoting the vehicle initiative safety. The pedestrian protection system can detect, identify and track the pedestrian from the video sequences, and also have an understanding and description of its behaviour, and thus activate effective warning. The accuracy of the object tracking directly determines the performance of the detection system.This paper has a research on pedestrian detection and tracking under the moving camera, against the problem of pedestrian safety in road traffic environment and based on particle filter theory. Described design process method of pedestrian tracking in detail based on particle filter algorithm and the concrete steps to achieve. The specific contents are as follows.(1) The paper presents a pedestrian detection approach based the histograms of oriented gradients (HOG) features of the pedestrians. Detailed HOG based pedestrian detection method is illustrated in literature [7]. HOG features extracted from the local area of the image and classifier is formed by using the look up table Gentle Adaboost algorithm.(2) To improve the effectiveness of pedestrian tracking, the HOG and color histogram characteristics are adopted to track pedestrian based on particle filter. Firstly, the pedestrian is detected using the HOG features to determine the initial target position. Then the target is tracked based on particle filter utilizing color histogram, during which the HOG is used to modify particle heavy weights and particle sampling. Experimental results verify the accurateness and efficiency of the proposed method.(3) The pedestrian often can’t be tracked accurately in the case of illumination changes and occlusions with the traditional algorithm. To improve the effectiveness of pedestrian tracking, particle filter (PF) is used to establish the pedestrian motion model and the color histograms is adopted to establish the object model for pedestrian tracking. Meanwhile. Scale Invariant Feature Transform (SIFT) features are used to correspond the region of interests across frames.(4)Finally, according to the foregoing findings, the pedestrian detection and tracking algorithm are organically combined to form a complete system through the matching technique. Pedestrian detection and tracking experiments are carried out under a few typical scenarios and compare the number of different particles for tracking time-consuming and tracking error.Experimental results indicate that the algorithm introduced could achieve effective recognition and tracking of pedestrians on the front of the vehicle. The pedestrian detection system can accurately depict the trajectory of pedestrian movement, and provide the basis for the analysis of its behavior.