针对光电装备的小目标检测与跟踪问题,总结分析了多类小目标检测与跟踪算法的原理和特点。从小目标的国际标准定义出发,小目标检测与跟踪算法主要分为两大类：空间滤波和时间滤波的方法。基于空间滤波的方法包括最大/中值滤波、门限检测法、二维最小均方滤波器、局部熵法、数学形态学等,基于时间滤波的方法包括多级假设检验、三维匹配滤波方法、高阶相关法、动态规划法等;通过对小目标特点的总结和各种算法的综合分析,从特定场景应用、时-空滤波结合、新理论新方法三个方面提出小目标检测与跟踪技术的发展方向,并总结指出小目标的检测与跟踪问题的本质是特征提取与滤波判决的问题。

For the small target detection and tracking with vision sensors, the principles and properties of small target detection and tracking algorithms are analyzed. From the international standard definition of small target, the small target detection and tracking algorithm includes two types: the method of spatial filtering and time filtering. The method of spatial filtering contains: max/median filtering, threshold detection, two-dimension least mean square, locally entropy, mathematical morphology, and etc. The method of time filter contains: multiple hypothesis test, 3-D matched filtering, high order correlation, dynamic programming, and etc. Through the conclusion of small target features and the comprehensive analysis of various algorithms, the development proposals of small target detection and tracking are proposed from three aspects of specific scene application, time-spatial filtering, new principle and new method. The essence of small target detection and tracking problem is the feature extraction and filtering judgment.