基于改进顶帽变换的红外弱小目标检测

Improved Top-hat Transform–based Algorithm for Infrared Dim and Small Target Detection

天空背景下的红外弱小目标检测技术较为成熟,但在近地复杂背景下,红外弱小目标的检测存在准确率不高、虚警目标多、实时性差的问题。针对以上问题,该文提出一种基于改进顶帽变换的红外弱小目标检测算法(OTHOLCM)。该算法采用基于改进顶帽变换的图像预处理算法(OTH),通过对不同灰度值的图像采取不同的策略针对性地处理图像,达到目标增强、背景抑制的效果。并在此基础上,采用基于改进多尺度局部对比度的红外弱小目标检测算法(OLCM),通过针对目标尺寸特点进行尺度设计,使得在保证算法实时性的基础上扩大目标尺寸检测范围。实验证明:OTHOLCM算法可以保证实时性并明显提高目标检测准确率、减少虚警目标数量。与3层模板局部差异度量算法(TTLDM)、基于边角感知的时空张量模型(ECASTT)等先进算法相比,OTHOLCM算法可使真阳性率分别提高近79%,61%,假阳性率分别降低近77%,73%,目标检测速度达到每秒25帧。

The technology for detecting infrared dim and small targets in the sky background is relatively mature.However,detecting these targets in near-ground complex backgrounds poses challenges such as low accuracy,high false alarm rates,and poor real-time performance.To address these problems,a novel algorithm for detecting infrared dim and small targets based on an improved top-hat transform,referred to as OTHOLCM,is proposed in this study.The algorithm uses an image preprocessing method,OTH,based on an improved top-hat transformation to enhance the target and suppress the background.Different strategies are employed to process images with different gray values.Additionally,the algorithm uses an infrared dim and small target detection technique,OLCM,based on improved multi-scale local contrast.The OLCM uses target size characteristics to expand the target detection range while ensuring real-time performance.Experimental results show that the OTHOLCM algorithm can guarantee good real-time performance,improve target detection accuracy,and reduce the number of false alarms.Compared with advanced algorithms such as the three-layer template local difference measurement algorithm and the edge and corner awareness-based spatial-temporal tensor,the OTHOLCM algorithm increases the actual positive rate by almost 79%and 61%,respectively.In addition,it reduces the false positive rate by nearly 77%and 73%,respectively.Moreover,the target detection speed reaches 25 frames per second.