Ground battlefield target tracking recapture nuclear filtering improved algorithm

The invention provides a ground battlefield target tracking recapture nuclear filtering improved algorithm based on a KCF algorithm. Judging whether the target is shielded or not according to the ratio of the feature response of the current detection frame to the average feature response value of all tracked video frames, predicting the position of the target by using Kalman filtering after the shielding occurs, and realizing target recapture when the predicted position is matched with the feature of the target before the shielding; if similar targets exist, position prediction and matching are carried out on the current target tracking frame, so that the problem of tracking errors is solved, the algorithm tracking stability is high, the algorithm can be deployed in embedded equipment, high real-time performance is achieved, and the target tracking requirement of a ground battlefield is met. 本发明以KCF算法为基础提出一种地面战场目标跟踪重捕获核滤波改进算法。以当前检测帧特征响应与已跟踪所有视频帧平均特征响应值比例判断目标是否遮挡，并于遮挡发生后使用卡尔曼滤波对目标进行位置预测，在预测位置匹配遮挡前的目标特征实现目标重捕获；