光电成像跟踪过程中，目标距离较近时，将在视场中呈现出扩展目标的特性，其成像大小将随着距离的减小迅速增大甚至溢出视场。此时传统的质心、形心、相关跟踪会出现跟踪点跳动或漂移的问题，严重时甚至出现跟踪丢失的现象。针对这一问题，提出了结合目标运动与主轴方向的目标头部跟踪方式，通过跟踪目标头部确保背景在视场内占据一定的比例，避免了目标充满视场而导致的分割失败或跟踪点在目标上滑动的情况。通过对目标的二值图像进行形态学滤波并计算头部的形心，进一步提高了头部跟踪点的稳定性。仿真测试与外场试验表明，该方法显著提高了跟踪的稳定性与可靠性。

In the process of tracking a target with photoelectric tracking system, the target represents the characteristics of extended target when it approaches the tracking device. The size of the target increases rapidly beyond the view of sight, as the distance between the target and the photoelectric tracking device gets nearer and nearer. The tracking point jumps or drifts away frequently when using conventional methods such as based on centroid, shape center and correlation tracking, and even lose the tracking. To solve the problem, a steady tracking point extraction method is proposed. It calculates the movement orientation and principal axis of the target. Through tracking the head of the target, it guarantees the background taking up some ratio in the view of sight and avoids failure of segmentation or tracking point drifts away on target when the target beyond the view of sight. The binary morphology method is used for filtering and the shape center of the target head is calculated, which increases the stability of the tracking point on the target head. The simulation test and the experimental results show that the method proposed increases the stability and reliability of the tracking remarkably.