Active tracking method based on high-speed dome camera

The present invention relates to an active tracking method based on a high-speed dome camera, in which by using an improved continuously adaptive mean shift tracking method, weighted fusion is performed on a probability distribution graph obtained by back projection of an HSV histogram with a probability distribution graph obtained by back projection of an image edge histogram, and the position of a tracking object is forecasted by combining with the Kalman filter algorithm, thereby improving the tracking robustness, and achieving that the pan tilt rotates with the motion of the tracking object in real time in the tracking process, to make the tracking object be within a designated range of an acquired image all the time and provide a good solution for some occasions requiring to track designated objects. 本发明涉及种基于高速球机的主动式跟踪方法，采用改进的连续自适应的均值漂移跟踪方法，将HSV颜色直方图反向投影得到的概率分布图，与图像边缘直方图反向投影得到的概率分布图进行加权融合，并结合卡尔曼滤波方法，预测跟踪目标的位置，从而提高跟踪的健壮性，并且能够实现云台在跟踪过程中实时地随跟踪目标运动而转动，始终让跟踪目标处于采集到的图像的指定范围内，为某些场合需要跟踪指定目标提供了种很好的解决方案。