Study Two Kinds Of Algorithms On Vehicle Tracking

Vehicle tracking is widely used in military and transport system. The key of vehicle tracking is correctly and timely. When there is part of occluded, light changed it can track goal vehicle correctly.In this thesis, the main content is vehicle tracking based on license plate and part of occluded vehicle tracking. When tracking vehicle, the license plate has unique character. It can help us know which vehicle we want. But the license plate is small it may be occluded, then we tracking vehicle based on feature points matching. To solve the problem of tracking which vehicle occluded effectively, a method based on improved ORB(improved Oriented FAST and Rotated BRIEF) feature matching is proposed.Firstly recognize license plateLremove noise from image then gray image, detect the edge of image. Morphology and the edge of image formed connected domain. Then combine feature of license plate we can get the region of license plate. Threshold and vertical projection make the character segmentation. To recognize the character we use contour matching. Then track the goal vehicle based on Kalman.When tracking license plate, it may be occluded due to the area is small. But the area of vehicle is big. It can help us track vehicle. In this course the goal vehicle may be occluded by other vehicles.To solve the real-time tracking problem of vehicle occluded effectively, a method based on improved ORB(improved Oriented FAST and Rotated BRIEF) feature matching is proposed. First FAST detects feature points then uses Laplace values of the pixel to eliminate false corners. Compared with ORB it improved rate and speed of matching. Improved FAST can get feature points quickly, BRIEF(Binary Robust Independent Elementary Features) can shorten time to describe the feature and reduce the space of storage to improve the speed of matching. It could meet the demand of real-time applications. Tests show that ORB can track the goal of vehicle when there is part of occludedLeven though there is noise or light changed.