一种基于改进卡尔曼滤波的camshift目标跟踪算法

Camshift Object Tracking Algorithm Based on Modified Kalman Filter

针对目标跟踪中的突变问题,本文提出一种基于改进卡尔曼预测的camshift(continuously adaptivemean shift)跟踪算法.本算法首先使用一种新的目标颜色模型,对传统目标模型进行改进,提高了目标跟踪的准确性和稳定性;同时为了更有效的预测目标位置,对卡尔曼滤波的一步预测值进行改进,并将修改后的卡尔曼预测算法融入camshift算法中,跟踪中增加采样率.实验表明,与传统camshift算法相比,该算法能够处理目标运动中发生突变的情况,实现对运动目标高精度的跟踪.

Aiming at the mutations problem in object tracking, this paper puts forward a new moving object tracking algorithm based on modified Kalman filter. This algorithm uses a new objective hue model, improves the objective model, and increases the object moving veracity and stability. Meanwhile in order to forecast the target site more effectively, we improve the step prediction value in the Kalman filter, make the new Kalman filter infuse camshift, and increase the sampling rate in tracking process. By the experiments, compared with traditional camshift, this algorithm can dispose the mutations problem, and bring about high-accuracy track- ing for object moving.