基于卡尔曼预测和滤波的视频目标跟踪

视频目标跟踪在军事和民用领域有着广泛的应用,其实现算法首选卡尔曼滤波算法,但是在应用卡尔曼滤波算法之前首先要获得目标当前时刻的观测信息,在全帧图像搜索并确定目标需要耗费大量的计算资源。针对这一问题,文章结合目标跟踪时的卡尔曼滤波算法,在每次完成目标状态的滤波估计后,进行卡尔曼一步预测,确定目标下一时刻可能的范围,避免下一时刻确定量测信息时对整幅图像的搜索,极大地减小计算量,提高目标跟踪的实时性和精度。计算机仿真实验验证文中所提方法的有效性。

Kalman Predictor and Filter Based Vision Target Tracking

Vision target tracking has wide applications in both military and civilian. The first choice of the realization algorithm is Kalman filter, but the measurement information of the current target should be achieved before the Kalman filter is applied. Large computation resource will be exhausted due to searching targets in the whole image. Aiming at this problem, this paper predicts the area where the target may appear at the next epoch by one step Kalman prediction, based on the position of the target estimated by Kalman filter at present, avoiding searching the whole image to find the target. It reduces the calculation burden and improves the real-time ability and precision. Computer simulations verify theavailability of the presented method.