一种基于目标与背景特征分离模型的高光谱目标检测修正算法

doi:10.3964/j.issn.1000-0593(2024)01-0283-09

摘要
参考文献
相关文章 (15)

全文: PDF (50424 KB)
输出: BibTeX | EndNote (RIS)

摘要：高光谱图像立方体数据可以提供成像场景中地物在可见光和近红外波长范围内的空间信息和地物属性诊断的光谱特征信息，在目标检测与识别方面拥有得天独厚的天然优势。然而，基于高光谱图像数据的目标检测也存在一定缺陷，如经典的高光谱目标检测算法仅利用光谱维度信息检测目标，检测模型要么对背景高维特征矩阵构建的准确度不足，要么对背景先验光谱特征的完备性要求较高，导致算法对不同复杂度的检测场景适应性不强。因此，基于计算复杂度较低、参数需求量较少且检测性能较为优异的经典多目标检测算法—多目标约束能量最小化（MCEM），提出了一种基于目标与背景环境特征分离模型的高光谱目标检测修正算法（R-MCEM）。首先，设计了一个与目标形状、尺寸相近的逐像元移动运算窗口，依次计算窗口中的每个像元与窗口内其他像元的光谱距离之和D1，像元与各类目标的光谱距离之和D2。其次，采用获得D1/D2最小值的像元替换窗口内的所有像元值。然后，自左向右、自上而下逐像元移动窗口，重复窗口内每一个像元与目标、背景像元的光谱距离运算，并确定窗口内与背景相似度最高、与目标相似度最低的像元。直到移动运算窗口遍历整个高光谱图像，大幅提升了基于目标与背景环境特征分离的背景高维特征矩阵准确度。分别设计了基于实测高光谱图像数据和模拟图像数据的修正检测算法性能验证试验，并采用三维操作特征曲线（3D ROC)结合目标与背景分离度（SDBT)开展修正算法的检测精度评估。试验结果表明，提出的修正算法有效减少了虚警率，提高了检测精度。基于实测数据的检测精度、目标与背景分离度由MCEM算法的0.937 7、0.57提升到R-MCEM的0.993 5、0.67，基于模拟数据的亚像元检测能力由MCEM的20%丰度提升到R-MCEM的15%丰度。

关键词：高光谱目标检测；目标与背景特征分离模型；3D ROC；SDBT

Abstract：Hyperspectral imagery cube Data can provide spatial information and diagnostic spectral characteristics, in the range of visible and near-infrared wavelength, about the attributes of materials in the scene, which contribute to the unique advantage of hyperspectral imagery for target detection. However, hyperspectral target detection has some shortcomings, such as the classical hyperspectral target detection algorithm only uses the information of spectral dimension to detect the target. The detection model either has insufficient accuracy for the construction of background high-dimensional f characteristics matrix or has high requirements for the completeness of background prior spectral characteristics, resulting in poor adaptability of the algorithm to different scenarios. Therefore, based on the classic multi-target detection algorithm-multiple target constrained energy minimization (MCEM), which has low computational complexity, fewer parameter requirements and better detection performance, this paper proposes a modified algorithm R-MCEM (revised MCEM) based on the separation model of target and background. First of all, this algorithm designs a pixel-by-pixel moving operation window that is similar to the shape and size of the target and sequentially calculates the spectral distances between each pixel and other pixels in the window, referred to as D1, and the spectral distances between the pixel and all targets,referred as D2. Next, all pixel values in the window are replaced with the pixel obtaining the minimum value of D1/D2. Then, move the window from left to right and from top to bottom and repeat the same calculation. Until the moving operation window traverses the entire hyperspectral image, all the interested targets in the hyperspectral image are eliminated as much as possible, and the accuracy of the background high-dimensional characteristics matrix is greatly improved. In this paper, the performance verification tests of the modified detection algorithm based on the true hyperspectral image data and the simulated image data are designed respectively, and the detection accuracy evaluation of the proposed algorithm is carried out by the three-dimensional receiver operation characteristics curve (3D ROC) combined with the Separation Degree Between Background and Target（SDBT）. The experimental results show that the proposed algorithm can effectively reduce the false alarm rate and improve the detection accuracy. The detection accuracy and SDBT based on the actual data are increased from 0.937 7 and 0.61 of the MCEM algorithm to 0.993 5 and 0.71 of R-MCEM, and the sub-pixel detection ability based on the simulated data is increased from 20% abundance of MCEM to 15% abundance of R-MCEM.

Key words：Hyperspectral target detection; Separation model between target and background; 3D ROC; SDBT

收稿日期: 2022-08-27 修订日期: 2022-11-15

中图分类号:

TP79

基金资助: 科工局课题（HDH59030303），装发领域基金项目(61402080404)资助

通讯作者: 邓贤明 E-mail: 15198121267@163.com

作者简介: 吴护林，1964年生，中国兵器工业第五九研究所研究员，中国南方工业集团有限公司首席科技专家
e-mail: zscqforwj_mail@163.com

引用本文:

吴护林，邓贤明, 张天才, 李忠盛，岑奕，汪家辉，熊杰，陈知华，林牧春. 一种基于目标与背景特征分离模型的高光谱目标检测修正算法[J]. 光谱学与光谱分析, 2024, 44(01): 283-291.
WU Hu-lin, DENG Xian-ming, ZHANG Tian-cai, LI Zhong-sheng, CEN Yi, WANG Jia-hui, XIONG Jie, CHEN Zhi-hua, LIN Mu-chun. A Revised Target Detection Algorithm Based on Feature Separation Model of Target and Background for Hyperspectral Imagery. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2024, 44(01): 283-291.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2024)01-0283-09 或 https://www.gpxygpfx.com/CN/Y2024/V44/I01/283

[1] ZHANG Bing，GAO Lian-ru（张兵，高连如）. Hyperspectral Image Classification and Target Detection（高光谱图像分类与目标探测）. Beijing: Science Press（北京: 科学出版社），2011.
[2] TONG Qing-xi, ZHANG Bing, ZHANG Li-fu（童庆禧，张兵，张立福）. National Remote Sensing Bulletin（遥感学报）, 2016, 20(5): 689.
[3] Xie W, Zhang X, Li Y, et al. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020, 13: 5887.
[4] Harsanyi Joseph C (Won Nobel for Game Theory). Detection and Classification of Subpixel Spectral Signatures in Hyperspectral Image Sequences. Baltimore: Department of Electrical Engineering, University of Maryland Baltimore County, 1993.
[5] Kraut S, Scharf L L. IEEE Transactions on Signal Processing, 1999, 47(9): 2538.
[6] Harsanyi J C, Chang C I. IEEE Transactions on Geoscience and Remote Sensing, 1994, 32(4): 779.
[7] PENG Bo，ZHANG Li-fu，ZHANG Peng，et al（彭波, 张立福, 张鹏，等）. National Remote Sensing Bulletin（遥感学报）, 2017，21(5): 739.
[8] Reed I S, Yu X. IEEE Transactions on Acoustics, Speech, and Signal Processing, 1990, 38(10): 1760.
[9] Yu X, Reed I S. IEEE Transactions on Signal Processing, 1993, 41(8): 2639.
[10] Qian D, Ren H. Proc SPIE, 2002, 4725: 501(doi: 10.1117/12.478783).
[11] Chang C I. IEEE Transactions on Geoscience and Remote Sensing, 2021, 60: 1(doi: 10.1109/TGRS.2021.3086768).
[12] Chang C I, Chiang S S. IEEE Transactions on Geoscience and Remote Sensing, 2002, 40(6): 1314.
[13] Hsueh M, Chang C I. IGARSS 2004. 2004 Geoscience and Remote Sensing Symposium, 2004, 5: 3222.
[14] Kwon H, Nasrabadi N M. IEEE Transactions on Geoscience & Remote Sensing, 2005, 43(2): 388.
[15] Lo E, Ingram J. Proc of SPIE, 2008, 6966: 696603.
[16] Zhai Y, Zhang L, Wang N, et al. IEEE Geoscience Remote Sensing Letter, 2016，13(8)：1059.
[17] Ren H, Qian D, Chang C I, et al. Comparison Between Constrained Energy Minimization Based Approaches for Hyperspectral Imagery. IEEE Workshop on Advances in Techniques for Analysis of Remotely Sensed Data. IEEE, 2004.
[18] DENG Xian-ming，MIAO Fang，ZHAI Yong-guang，et al（邓贤明，苗放，翟涌光，等）. Acta Scientiarum Naturalium Universitatis SunYatseni[中山大学学报(自然科学版)], 2017, 56(1)：151.
[19] Chang C I. IEEE Transactions on Geoscience and Remote Sensing, 2021, 59(6)：5131.
[20] ZHAO Chun-hui，LI Xiao-hui，WANG Yu-lei（赵春晖, 李晓慧, 王玉磊）. Journal of Electronic Measurement and Instrumentation（电子测量与仪器学报）, 2014, 8: 803.
[21] Herweg J A, Kerekes J P, Weatherbee O, et al. SpecTIR Hyperspectral Airborne Rochester Experimental Data Collection Campaign. Proc. SPIE，2012，8390.