﻿ 基于频域残差及局部协方差的红外弱小目标检测
 舰船科学技术  2023, Vol. 45 Issue (23): 139-144    DOI: 10.3404/j.issn.1672-7649.2023.23.024 PDF

Infrared dim small target detection based on spectral residuals and local covariance
LI Dong
Jiangsu Automation Research Institute, Lianyungang 222061, China
Abstract: In complex scenes, the infrared small target can be lost in the background and noise easily, due to features of small size, weak signal, lack of texture, resulting in high false alarm detection rate, complex algorithm, large amount of calculation and other problems. We present an infrared dim small target detection method based on spectral residuals and local covariance. Firstly, the saliency map can be obtained by calculating the spectral residuals of the original infrared images, which can obtain the possible region of the target. Secondly, the local covariance detection method can be used for identification in this area. Finally, the small target can be detected by adaptive threshold segmentation method.Experimental results indicate that compared with traditional detection algorithms, the proposed algorithm can effectively suppress background and noise in different scenes, accurately detect targets, and meet the real-time requirements.
Key words: target detection     spectral residual     local covariance detection     detection of significance     adaptive threshold
0 引　言

1 算法原理

 图 1 红外弱小目标检测流程原理图 Fig. 1 Schematic of infrared small target detection process
1.1 基于频域残差的显著区域提取

 $A(u,v) = |FFT(I(x,y))| ，$ (1)
 $P(u,v) = Pha(FFT(I(x,y))) ，$ (2)
 $R(u,v) = \log (A(u,v)) - h(u,v)*\log (A(u,v))，$ (3)
 $S(x,y) = G(x,y)*IFFT({(\exp (R(u,v) + iP(u,v)))^2} 。$ (4)

 图 2 红外图像及相应的显著区域提取结果 Fig. 2 Infrared images and corresponding salient regions extraction results

1.2 基于局部协方差的目标检测

 $\overline {Dis} = {{\left( {\sum\limits_{m = 1}^5 {\sum\limits_{n = 1}^5 {\sqrt {({{(m - \overline m )}^2} + {{(n - \overline n )}^2})} } } } \right)} \mathord{\left/ {\vphantom {{\left( {\sum\limits_{m = 1}^5 {\sum\limits_{n = 1}^5 {\sqrt {({{(m - \overline m )}^2} + {{(n - \overline n )}^2})} } } } \right)} {\left( {N - 1} \right)}}} \right. } {\left( {N - 1} \right)}}，$ (5)
 $\overline {GD} = {{\left( {\sum\limits_{m = 1}^5 {\sum\limits_{n = 1}^5 {\left( {{I_o} - {I_{mn}}} \right)} } } \right)} \mathord{\left/ {\vphantom {{\left( {\sum\limits_{m = 1}^5 {\sum\limits_{n = 1}^5 {\left( {{I_o} - {I_{mn}}} \right)} } } \right)} {\left( {N - 1} \right)}}} \right. } {\left( {N - 1} \right)}} ，$ (6)
 $\begin{split} Cov(Dis,GD) = &- \left( \sum\limits_{m = 1}^5 \sum\limits_{n = 1}^5 \left( \sqrt {{{(m - \overline m )}^2} + {{(n - \overline n )}^2}} - \overline {Dis} \right) \times\right.\\ &\left( {{I_o} - {I_{mn}} - \overline {GD} } \right) \Biggr) \Biggr/ {(N - 1)}，\\[-20pt]\end{split}$ (7)
 $\begin{split}& Cov\_w(Dis,GD) =\\ & \left\{ \begin{array}{*{20}{c}} \dfrac{{GD}}{{Dis}}Cov(Dis,GD),& Cov(Dis,GD) \geqslant 0 ，\\ 0,& {\rm{otherwise}}。\end{array} \right.\end{split}$ (8)

 图 3 协方差显著图及其三维图 Fig. 3 Covariance Salient maps and gray response maps
1.3 红外弱小目标提取

 ${T_h} = {\mu _s} + k \times {\sigma _s} 。$ (9)

 图 4 图像检测结果 Fig. 4 Detection results of image
2 实验及分析 2.1 实验数据与评估标准

 图 5 四种测试场景的真实图像 Fig. 5 Real images under four scenes

 $SCRG = \frac{{SC{R_{out}}}}{{SC{R_{in}}}}。$ (10)

 $SCR = \frac{{T - B}}{\delta } 。$ (11)

 $BSF = \frac{{{\sigma _{{\rm{in}}}}}}{{{\sigma _{{\rm{out}}}}}} 。$ (12)

2.2 实验结果与对比分析

 图 6 四种测试场景下不同算法的检测结果 Fig. 6 The results of different algorithms under four scenes

3 结　语

 [1] CHEN C P, LI H, WEI Y T, et al. A local contrast method for small infrared target detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2014, 52(1): 574-581. DOI:10.1109/TGRS.2013.2242477 [2] WU L, MA Y, FAN F, et al. A double neighborhood gradient method for infrared small target detection[J]. IEEE Geoscuence and Remote Sensing Letters, 2021, 18(8): 1476-1480. DOI:10.1109/LGRS.2020.3003267 [3] WANG X, LV G F, XU L Z. Infrared dim target detection based on visual attention[J]. Infrared Physics & Technology, 2012, 55(6): 513-512. [4] HOU X D, ZHANG L Q. Saliency detection: a spectral residual approach[J]. IEEE Conference on Computer Vision and Pattern Recognition, 2007, 800: 1-8. [5] 黄敏, 鲍苏苏, 邱文超. 基于可见光下双目视觉的手术导航研究与仿真[J]. 机器人, 2014, 36(4): 461-468,476. HUANG M, BAO S S, QIU W C. Study and simulation of surgical navigation based on binocular vision under visible light[J]. Robot, 2014, 36(4): 461-468,476. DOI:10.13973/j.cnki.robot.2014.0461 [6] NIE J Y, QU S C, WEI Y T, et al. An infrared small target detection method based on multiscale local homogeneity measure[J]. Infrared Physics & Technology, 2018, 90: 186-194. [7] GU Y F, WANG C, LIU B X, et al. A kernel-based nonparametric regression method for clutter removal in infrared small-target detection applications[J]. IEEE Geoscuence and Remote Sensing Letters, 2010, 7(3): 469-473. DOI:10.1109/LGRS.2009.2039192 [8] TOM V T, PELI T, LEUNG M, et al. Morphology based algorithm for point target detection in infrared backgrounds[J]. Proceeding. SPIE, 1993, 1954: 25-32. DOI:10.1117/12.157777 [9] 李凡, 刘上乾, 秦翰林. 自适应双边滤波红外弱小目标检测方法[J]. 光子学报, 2010, 39(6): 1129-1131. LI F, LIU S Q, QING H L. Dim infrared targets detection based on adaptive bilateral filtering[J]. Acta Photonica Sinica, 2010, 39(6): 1129-1131. DOI:10.3788/gzxb20103906.1129 [10] 张晓露, 李玲, 辛云宏. 基于小波变换的自适应多模红外小目标检测[J]. 激光与红外, 2017, 47(5): 647-652. ZHANG X L, LI L, XIN Y H. Adaptive multi-mode infrared small target detection based on wavelet transform[J]. Laser & Infrared, 2017, 47(5): 647-652.