Multiple-kernel based vehicle tracking using 3-D deformable model and license plate self-similarity

摘要: In this paper, we propose a novel vehicle tracking system under surveillance camera. The proposed tracks vehicles by using constrained multiple-kernel, facilitated with Kalman filtering, to continuously update the position and orientation of moving vehicles. To further reliably track partial occlusion or even total occlusion, our algorithm also systematically builds 3-D model, from which license plate region is identified self-similarity descriptor used for low-resolution matching. Experimental results have shown favorable performance system, can successfully serious while maintaining knowledge geometry tracked