跟踪窗口尺寸自适应调整的粒子滤波跟踪算法

当目标尺度发生变化时,传统的粒子滤波跟踪算法的跟踪窗口尺寸不变,在目标尺寸变化较大时容易丢失跟踪目标.针对这一问题,提出了一种跟踪窗口自适应调整的粒子滤波跟踪方法.该方法依据运动目标区域内粒子到目标中心点的平均距离与目标尺寸的关系,建立跟踪窗口尺寸的数学模型.在两种目标模型上对所建立的数学模型进行了仿真验证.实验结果表明,当目标尺度发生变化时,跟踪窗口能够很好的随目标的尺寸变化而自适应地连续调整,改进后的算法在目标尺寸变化率很大时仍能够稳定跟踪目标.

Adaptive Window Object Tracking for Particle Filter

This paper proposes an adaptive window object tracking method based on particle filter. It figures out size-changes of moving objects during target tracking. This method firstly establishes the gray histogram statistics of the observation model, using Bhattacharyya coefficient similarity for forecasting center position of target. Then it sets up a mathematical model for adjusting object size with a target window according to the average distance changes. The average distance is the particles in the moving target window to the target center. Experimental results indicate that the algorithm of tracking window can adaptively change with size-changes of moving objects, and the target results are good.