Real-time Tracking of Underwater Moving Target

Swimming pool is a dynamic scene with great changes. It is affected by bubbles, light intensity, wave and other factors. In view of the features of the swimming pool, the underwater target tracking algorithm based on the combination of Mean Shift and Particle Filter is proposed in this paper. At first, according to the kernel function and target model in Mean Shift, the underwater moving target model is build based on RGB histogram. Then particle filter algorithm is used to track moving target. In the process of tracking, the distributions of the particles get closer to the true state of target by using the convergence of Mean Shift on particles. The simulation results show that the algorithm proposed in this paper achieves real-time and stable moving target tracking in the complex surroundings of swimming pool.