融合视觉显著性和局部熵的红外弱小目标检测

Infrared dim small target detection based on visual saliency and local entropy

针对红外图像中弱小目标检测虚警率高、实时性差的问题,提出了一种基于视觉显著性和局部熵的红外弱小目标检测方法。该方法将红外弱小目标的检测问题由粗到精分步实现,首先利用融合局部熵的方法提取包含目标的感兴趣区域,对红外弱小目标实现粗定位。然后再利用改进的视觉显著性检测方法在感兴趣区域计算局部对比度,获得感兴趣区域的显著图。最后利用阈值法分割显著图像提取红外弱小目标,实现红外弱小目标的检测。通过与TOPHAT算法及LCM算法进行对比试验,验证了该方法在检测性能上优于TOPHAT算法以及LCM算法,虚警率分别下降了62.5%和33.3%;检测实时性方面,算法耗时为LCM的38.6%。该方法能够实现复杂背景下红外弱小目标的准确检测,在一定程度上解决了目标检测虚警率高、实时性差的问题。

To improve the high false-alarm rate and poor real-time capability in detecting infrared small dim targets,a novel algorithm based on visual saliency and local entropy is proposed in this paper.This method solves the problem from coarse to fine detecting of small targets.First,a local entropy method is used to ob-tain the region of interest.Then,an improved visual saliency method is used to calculate local contrast.Fi-nally,a threshold segmentation method is used to extract dim infrared small targets.The method is verified using a contrast test with TOPHAT and LCM,and the results show that the performance of this method pre-cedes the TOPHAT algorithm and LCM algorithm.The false alarm rate by this method decreases to 62.5%and 33.3%compared with the other two algorithms,and the time cost decrease to 38.6%of that of LCM.The method can achieve accurate detection of infrared dim and small targets in a complicated environment,solv-ing the high false alarm rate and poor real-time capability issues to some extent.