基于频域显著性分析的红外小目标检测算法  

Small Infrared Target Detection Algorithm Based on Frequency Domain Saliency Analysis

作　　者：孙泽军[1] 丁萌[1,2] 王洁 王跃东[1] 

机构地区：[1]南京航空航天大学,南京211106 [2]光电控制技术重点实验室,河南洛阳471023

出　　处：《电光与控制》2017年第10期1-6,共6页Electronics Optics & Control

基　　金：国家自然科学基金(61203170);航空科学基金(20155152041)

摘　　要：针对复杂背景下的红外小目标检测问题,提出一种基于频域显著性分析的小目标检测算法。算法利用红外图像中目标在频域内相较于背景更加显著的特点,通过频域显著性计算得到红外图像的显著图,消除部分背景杂波干扰,然后通过自适应阈值分割显著图,提取出感兴趣区域,进一步在感兴趣区域中计算多尺度窗口的显著度,从而完成小目标的检测。从理论上分析了算法的有效性,并利用典型的红外图像进行了实验,实验结果表明,所提算法能够很好地完成低信噪比条件下的红外小目标检测。与其他方法相比,在保证目标检测准确率的前提下,所提算法简单有效、复杂度低、计算效率高,满足实时性要求。To solve the problem of small infrared target detection under complex backgrounds, an effective algorithm based on saliency analysis in frequency domain is proposed. The infrared target is usually more significant than backgrounds in spectral domain. Thus, we can obtain the saliency map of the infrared image through some operations in frequency domain. In this way, target signal is enhanced and background clutter is suppressed. Then, an adaptive threshold is adopted to segment the saliency map and extract the region of interest. At last, the saliency of windows in the region of interest is measured to predict the exact position of the small targets. Theoretical analysis was made to the effectiveness of the algorithm. And to validate our algorithm, we conducted experiments on some typical infrared images that contain small targets. Experimental evaluation results show that our method can implement infrared small target detection under low signal-to-noise ratio fairly well, and the method is simple, effective, and can satisfy the real-time requirement while guaranteeing the detection accuracy.

关 键 词：目标检测 红外小目标 显著性 感兴趣区域 窗口显著度 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]