多波段红外弱小目标检测算法研究

Study on Dim Target Detection Algorithm Based on Multi-band Infrared Images

针对多波段红外弱小目标检测问题,建立多波段红外图像模型,从集中式和分布式检测处理两个角度,推导在目标和背景信息完全已知的理想条件下最优融合检测器形式,给出检测算法的性能上界。讨论在目标和背景信息不同程度缺失情况下的广义似然比(GLRT)融合检测器形式,提出一种实用的恒虚警分布式融合检测方法,给出可实现融合检测器的检测性能公式。在此基础上,对各类融合检测算法性能进行仿真对比分析,给出分析结论。本文对多波段红外弱小目标检测问题的研究较为系统、全面,得出的结论对不同工作条件下弱小目标融合检测器的选择设计具有指导作用。

The problem of dim target detection based on infrared multi-band images is addressed. After modeling the infrared multi-band image data,the optimal fusion detector in the ideal case,in which all information of the target and noise is known,is derived from the central and distributed detection aspects respectively. The performance of the optimal detector,which gives an upper bound of detection performance for all potential detectors,is analyzed. Then the GLRT based fusion detector is discussed in different cases in which the information of the target and noise is partially or entirely unknown and a utility CFAR-based distributed detector is proposed. The detection performance formulas of these detectors are derived. Based on these results,we compare and analyze the performances of all kinds of detectors through simulation,some conclusions are given. The problem of dim target detection based on infrared multi-band images is studied systematically and roundly in this paper,the conclusions presented are useful and instructional for designing dim target fusion detector under different situations.