复杂背景可见光图像中弱小目标探测的新算法

针对具有复杂背景的可见光图像中弱小目标很难探测这一问题,研究了一种面形态学空间滤波器可以在单帧图像中检测出比局部背景灰度值高出 10%以上的候选小目标;在经过空间滤波和帧相关滤波后的多帧图像中提取出每一个候选小目标的一组特征矢量;特征矢量的融合算法可以排列出多个小目标的威胁次序。仿真试验表明该算法可以探测出多个与背景灰度接近的可见光弱小目标,并可以对其威胁次序进行排序。

A new algorithm for small and dim target detection of visible image under heavy clutters

In view of the difficulty to detect small and dim targets in a group of visible images under heavy clutters, a spatial filter is designed using an area morphology algorithm which can find out the doubtful targets with more than 10% gray scale than the local background in single image. A group of characteristic vectors for each doubtful small target can extracted from multi-frame image after spatial filtering and frame correlation filtering. The fusion algorithm for characteristic vector can list threat sequence of multiple small targets. The simulation shows that the algorithm can detect visible small and dim targets with similar background gray scale and list their threat sequences.