Moving vehicle target tracking based on improved Mean shift algorithm

This article proposes a tracking algorithm based on mean shift and particle filter according to the characteristics of moving vehicles. Firstly, the algorithms establish auto target motion model that the core is RGB color histogram. Secondly, using the Bhattacharyya distance measures similarity of particle region and the target model, and updating the weights of particles according to the similarity. Lastly, the measure improves the estimated position of the particle by using mean shift algorithm, which makes the candidate regions of these particles can be more close to the true target position. Experimental results show that the algorithm has good real-time performance and robustness, and can achieve stable tracking of target vehicle.