| Because of the rapid growth of quantities of vehicles and road accidents, how to reduce road accidents has become the focus of attention. Particularly in backing, the rear blind spots of vehicles, together with the uncertainty of movement direction of moving targets, it is prone to collision between moving object and ego vehicle. Therefore, give immediately warning before possible collisions between moving objects and ego vehicle to offer sufficient protection to them, has received considerable attention in the field of Driver Assistance Systems. Dynamic scene, the diversity of moving object, including non-rigid body pedestrians and rigid body vehicles, as well as the weather, light and other factors make the moving targets detection and tracking very challenging.In this paper, we get the external environment information by means of on-board fish-eye camera. From the real practical point of view, we found moving target detection and tracking system.In the moving target detection system, firstly, this paper proposes a hierarchical ego-motion compensation method to eliminate the impact of camera motion. Secondly, processes the difference image with morphological processing, laying a foundation for segmentation and validation. Then, presents a difference grid based motion segmentation method, which is simple and effective. Finally, proposes a moving target validation algorithm based on the camera imaging theory.In the moving object tracking system, this paper presents an improved particle filter algorithm for multi-target tracking based on the difference image. The algorithm uses the difference value as tracking feature, which intuitiveness, calculation simple, and can cover translation, rotation, and scaling motions.The proposed moving object detection and tracking algorithms were tested with a large number of different scenarios in the test sets, and the results shown that the proposed algorithms were robust and exact. |
PDF Full Text Request