Multifeature Based Visual Tracking for Surveillance

Computer Engineering

This project attempts a robust vehicle tracking technique considering all the problems that would otherwise disturb the vehicle tracking. Like the variation in the pose of the vehicle, change in illumination in the video, occlusion as of any material coming in between the line of focus and the camera. There were various methods that were proposed in the literature before but sparse representation has been a more superior method. In this paper we build up an algorithm to optimize the weight updation involved in multi-kernel fusion on visual tracking system. This makes this algorithm to be adaptive and optimized. Particle Swarm Optimization (PSO) algorithm is introduced in order populate the weight value and update it in every iteration in order to attain the objective function which will provide us the sparsest realization of the feature from the frame. The previous work had put light on the multi-kernel fusion alone but we improve that to be a multi-objective optimization using Particle Swarm Optimization algorithm. The thought generated must be tested with the software tools like Mat lab to validate the efficiency.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.