局部最大熵的红外小目标快速检测方法

Infrared small target detection method based on local maximum entropy

近年来,红外弱小目标检测与跟踪技术在信息处理技术领域被广泛研究。航空报警、导弹拦截、近地面监视等多个领域均有较为出色的应用。其中,在不同的环境中如何有效地测量目标是研究人员面临的一项主要挑战。为了解决红外图像中的杂波干扰过多与低目标信噪比等一系列问题,提出了一种基于局部最大熵的红外小目标检测方法。为了提升算法的鲁棒性,提出一种自适应阈值,求出局部最大熵对应的灰度值作为阈值与目标与背景间的灰度差作比较,经过三轮不同判断,找到目标点。最后通过抑制非极大值抑制来去除非目标点。试验结果表明,在复杂多变的背景或低信噪比的情况下,仍具有有效的检测性能,通过与经典top-hat算法的对比实验,观测到无论是在检测时间还是检测准确率,所提出的方法均高于top-hat算法,其中top-hat算法的DR值为88.57,提出的方法的DR值为92.5,top-hat算法的FA值为0.1857,所提出的方法为0。

In recent years,the infrared weak and small target detection and tracking technology has been widely studied in the field of information processing technology.Aviation alarm,missile interception,near-ground surveillance and other fields have excellent applications.Among them,how to effectively measure targets in different environments is a major challenge for the researchers.In order to solve a series of problems such as excessive clutter interference and low target signal-to-noise ratio in infrared images,we propose a method for small infrared target detection based on local maximum entropy.In order to improve the algorithm robustness,we propose an adaptive threshold,find the gray value corresponding to local maximum entropy and compare it with gray difference between the target and the background,and go through three different judgments.Finally,the non-target points are removed by suppressing non-maximum suppression.The experimental results show that,in the case of complex and variable background or low signal-to-noise ratio,it still has effective detection performance,and its overall detection speed is higher than other linear filtering algorithms.Through a comparison experiment with the classic top-hat algorithm,it is observed that the proposed method is higher than the top-hat algorithm in both detection time and detection accuracy.The DR value of the top-hat algorithm is 88.57,the DR value of proposed method is 92.5.The FA value of the top-hat algorithm is 0.1857,and the proposed method is 0.