提出了一种基于矩特征和特征光流的运动目标跟踪方法。首先进行角点特征提取,按照提出的基于矩特征的局部范围内匹配角点的策略,完成了序列图像的角点匹配;然后,按照本文提出的光流聚类准则完成了两个图像目标的聚类。仿真实验表明,本文算法在减少计算量的同时可提高跟踪精度,且可克服目标做小角度旋转时的失跟问题。
In this paper, a moving target tracking method based on moment character and feature-optical-flow is proposed. The corner feature is firstly extracted for object matching, the matching strategy refers to Hu moment local matching method, then, a new feature-optical-flow clustering algorithm is presented to finish the clustering of two objects. Simulation results show that this method can improve the tacking accuracy while decreasing the computational burden. It also overcomes the track-lost problem when the object has a little deflexion.