双模式相关技术研究与实现

文中研究的相关跟踪技术主要应用于飞航导弹的末制导。现阶段，激光制导技术、GPS制导技术、合成孔径雷达（SAR）制导技术在我国的实际应用还不成熟，传统的电视、红外制导技术仍然具有很强的生命力，因而，基于可见光的视频相关跟踪技术具有重要的研究价值。本文主要论述的两种相关算法是：多灰度点相关（MPC）、区域模板相关（RTC）。其中多点相关（MPC）算法的跟踪灵敏度高，定位精度好，硬件实现比较方便，实时性能好；区域模板相关算法（RTC），在图像的匹配过程中，不仅考虑了目标区域的灰度特征，而且兼顾了区域里多灰度层次的位置特征、面积特征，算法具有很好的鲁棒性。文中深入研究了两种相关跟踪算法，并针对它们在实际应用中的不足，提出了有效的改善措施。最后，本文对两种相关跟踪算法进行了初步融合，一是：通过粗匹配、精匹配过程来选取目标跟踪点；二是：提出了一种度量模板更新的能量准则函数。大量的仿真实验结果表明：改进后的两种相关跟踪技术可以较好地完成一些复杂背景下的目标跟踪任务，两种算法的有效结合又进一步提高了目标跟踪的稳定性能和可靠性能。本文研究的一些相关跟踪技术已经运用到实际工程项目中。

A correlation algorithm studied in this paper is mainly applied to terminal air to ground missile guidance. Now the practical application of laser、GPS、SAR terminal missile guidance in China is not mature, and conventional E.O、infrared missile guidance is still of great potential. Therefore correlation algorithm based on video signal is of great study value. Two kinds of correlation algorithm studied in this paper are multiple point correlation (MPC) and region template correlation (RTC). MPC algorithm has high tracking sensitivity and good locating accuracy. Its hardware can be simply constructed so that it has good real-time performance. RTC algorithm not only considers the intensity feature of target region, but also considers multi-layer location feature and area feature. So it is a rather robust algorithm. To deal with the problems existed in practical application, these two kinds of algorithm are further explored and some efficient measure is put forward. Finally, these two kinds of algorithm have been combined in this paper, first: acquiring a tracking point through raw and fine image registration; second: a kind of energy function proposed to measure template updating. Many experiment results show that under complex background and foreground, two kinds of algorithm improved in the paper can track target more steadily. A hybrid algorithm has further upgraded target tracking probability and precision. Now, some correlation tracking technology has already been applied to practical engineering project.