【摘要】 为了跟踪复杂情况下运动速度较快的可变形目标的活动轮廓,提出一种变形峰值跟踪卡尔曼粒子滤波算法.该算法通过使用卡尔曼滤波器预测目标活动轮廓的全局平移速度,并将目标状态空间分为有效基空间(维数很小)和剩余基空间,对有效基空间进行重要性采样,而对剩余基空间使用峰值跟踪,从而实现对目标轮廓的跟踪.实验结果表明,该算法可有效跟踪运动速度较快的可变形目标轮廓.更多还原

【Abstract】 An algorithm for deforming Kalman particle filter with mode tracker was proposed to tracking the active contour of fast moving and deforming objects again complex background.In this algorithm,the global moving velocity of the active contour of an object was predicted and the state space was split into the effective basis space(whose number of the dimensions is small) and the residual basis space;important sampling was done on the effective basis and mode tracking on residual basis to track the contour of the object.The experiments show that the algorithm can efficiently track fast moving and deforming objects.更多还原