Object Tracking Using Meanshift Algorithm Combined with Kalman Filter on Robotic Fish
暂无分享,去创建一个
This paper investigates and proposes an improved Meanshift algorithm combined with Kalman Filter aiming at the shortcomings of the Meanshift algorithm theory as well as obvious limitations of a target tracking for the independent visual robotic fish being affected by the fluctuation of the water wave. First, this new algorithm makes use of Kalman filter to obtain the initial position of the Meanshift algorithm. Then, adjust the bandwidth of the kernel function adaptively in the Meanshift tracking algorithm and use the Meanshift algorithm to obtain the position of the tracking target. Finally, we conduct a real-time tracing experiment on the independent visual robotic fish tracking a moving ball. Experimental results show that: compared with the traditional Meanshift algorithm, the improved algorithm tracks the target more accurately and the trajectory of the tracking target is more continuous. Furthermore, it reduces the number of iterations, make the algorithm run faster and improve the real - time in tracking.
[1] R. E. Kalman,et al. A New Approach to Linear Filtering and Prediction Problems , 2002 .
[2] T. Başar,et al. A New Approach to Linear Filtering and Prediction Problems , 2001 .
[3] Larry D. Hostetler,et al. The estimation of the gradient of a density function, with applications in pattern recognition , 1975, IEEE Trans. Inf. Theory.
[4] Dorin Comaniciu,et al. Kernel-Based Object Tracking , 2003, IEEE Trans. Pattern Anal. Mach. Intell..