基于粗糙集的红外弱小目标检测方法

提出了利用粗糙集理论处理不完整信息的能力对红外弱小目标的运动轨迹进行辨识的检测方法。首先利用形态学滤波算法，分离出每帧图像中候选目标并提取其特征属性。把提取到的特征属性作为条件属性，把每帧图像中候选目标作为个体，构成知识决策系统。通过对决策表进行约简，得到决策系统的最小决策算法，并利用序列图像中目标运动的连续性和轨迹一致性来实现小目标的识别。实验表明，该方法能够有效解决目标跟踪过程中目标短暂丢失以及重现的问题，实现弱小目标的稳健检测和跟踪。

Rough set based dim small target detection method in infrared image sequences

A novel method that makes use of rough set theory to detect moving small target with low SNR in infrared image sequences is presented. First, Morphology Top-Hat operator is used to separate candidate small targets and extract their attributes. Then, the candidate targets are regarded as unit, and corresponding attributes as condition attributes. Subsequently the decision system of target′s tracks recognition is constructed. After decision table reduction, the minimal decision rules can be obtained. According to the motion continuity and the motion trajectory consistency in multi-frame successive image, the reduced minimal decision rules can discriminate the real target from the candidates. Experimental results show that the method can effectively solve the problems of targets missing and appearing again during the detecting and tracking procedure.