基于偏振光谱融合的镜面反射去除方法及其在植被健康监测中的应用

doi:10.3964/j.issn.1000-0593(2023)11-3607-08

摘要
参考文献
相关文章 (15)

全文: PDF (7580 KB)
输出: BibTeX | EndNote (RIS)

摘要：植被遥感监测已广泛应用于各个领域，如农作物病虫害监测、森林覆盖度监测、植被生长状况监测等。监测植物叶绿素含量变化对于了解植物长势、监测植被病虫害乃至监测植被对全球气候变化反馈都有着重要的意义，然而这些监测时常受到叶片镜面反射的干扰，导致叶绿素含量反演精度降低。旨在消除植物健康状况遥感监测中的镜面反射干扰，搭建一套偏振多光谱成像系统，提出一种镜面去除指数（SRRI），并提出一种利用目标的漫反射和镜面反射的光谱和偏振特性来检测植物的融合算法，植物的SRRI、线偏振度（DoLP）和偏振角（AOP）均在融合框架内计算，以消除植物镜面反射的干扰并提高植物健康状态检测精度。此外，基于SRRI、DoLP和AOP的融合算法计算了一种偏振融合镜面去除指数（PFSRRI）。对相对叶绿素含量（SPAD）、比值植被指数（SR）、归一化植被指数（NDVI）、SRRI_SR、SRRI_NDVI、PFSRRI_SR和PFSRRI_NDVI进行了相关性分析，以了解它们消除镜面反射干扰的能力。结果表明，SR与SPAD (R²=0.012 8)、NDVI与SPAD (R²=0.007 5)的相关性最差，表明SR和NDVI对镜面反射的敏感性最高，SRRI_SR与SPAD (R²=0.818)、SRRI_NDVI与SPAD(R²=0.889)有很好的相关性，而PFSRRI_SR与SPAD(R²=0.955)、PFSRRI_NDVI与SPAD(R²=0.948)的相关性最好，从而突出了PFSRRI在消除镜面反射干扰并检测植物健康状态中的潜力。PFSRRI_SR和PFSRRI_NDVI三维散点图显示了对植物不同健康程度具有良好的辨别能力，具有较高的敏感性和特异性值，通过曲面的颜色和趋势的变化可以很直观地看到植被健康状态的变化趋势和分类状况。其中，PFSRRI_SR的镜面叶片和胁迫1级叶片的分类敏感性（Se）值为100%和特异性（Sp）值为100%，PFSRRI_NDVI的镜面叶片和胁迫1级叶片的分类敏感性（Se）值为98%和特异性（Sp）值为100%，表明PFSRRI_SR和PFSRRI_NDVI在去除镜面干扰后的优秀的检测效果。综上所述，该方法能有效地消除镜面干扰，提高植被健康状况检测精度。

关键词：遥感；镜面反射；图像融合；植被指数；植被健康监测

Abstract：Vegetation remote sensing monitoring has been widely used in various fields, such as crop disease and insect pest monitoring, forest coverage monitoring, and vegetation growth monitoring. Monitoring changes in plant chlorophyll content is of great significance for understanding plant growth, monitoring vegetation pests and diseases, and even monitoring vegetation feedback on global climate change. However, these monitoring are often disturbed by the specular reflection of leaves, which reduces the inversion accuracy of chlorophyll content. This paper aims to eliminate the specular reflection interference in remote sensing monitoring of plant health, a polarization multispectral imaging system was established, a specular reflection removal index (SRRI) was proposed. A fusion algorithm was proposed to detect plants based on the spectral and polarization characteristics of diffuse and specular reflection of vegetation. SRRI, degree of linear polarization (DoLP) and angle of polarization (AOP) are all calculated in the fusion algorithm to eliminate the interference of specular reflection and improve the accuracy of plant health status detection. In addition, a fusion algorithm based on SRRI, DoLP and AOP calculates a polarization fusion specular reflection removal index (PFSRRI). Correlation analysis was performed on relative chlorophyll content (SPAD), ratio vegetation index (SR), normalized vegetation index (NDVI), SRRI_SR, SRRI_NDVI, PFSRRI_SR and PFSRRI_NDVI to understand their ability to eliminate specular reflection interference. The results showed that SR and SPAD (R²=0.012 8) and NDVI and SPAD (R²=0.007 5) had the worst correlation, indicating that SR and NDVI had the highest sensitivity to mirror reflection. SRRI_SR and SPAD (R²=0.818), and SRRI_NDVI and SPAD (R²=0.889) had a good correlation. The correlation between PFSRRI_SR and SPAD (R²=0.955) and PFSRRI_NDVI and SPAD (R²=0.948) was the best, which highlighted the potential of PFSRRI in eliminating mirror reflection interference and detecting plant health status. PFSRRI_SR and PFSRRI_NDVI 3d scatter plots show good discrimination ability for different health degrees of plants, with high sensitivity and specificity. The variation trend and classification status of vegetation health state can be intuitively seen through the color and trend of the surfaces. Among them, the sensitivity and specificity of PFSRRI_SR to classify specular leaves from stress level-1 was 100% and 100%, and the sensitivity and specificity of PFSRRI_NDVI to classify specular leaves from stress level-1 was 98% and 100%, indicating the excellent detection performance of PFSRR_SR and PFSRRI_NDVI after removing specular interference. In summary, this method can effectively eliminate specular reflection interference and improve the detection accuracy of vegetation health status.

Key words：Remote sensing; Specular reflection; Image fusion; Vegetation Index; Vegetation health monitoring

收稿日期: 2022-05-16 修订日期: 2022-09-13

中图分类号:

TP79

基金资助: 国家自然科学基金项目(62175144)资助

通讯作者: 王驰 E-mail: wangchi@shu.edu.cn

作者简介: 李思远，1994年生，上海大学机电工程与自动化学院博士研究生 e-mail: LSYYJ@shu.edu.cn

引用本文:

李思远，焦健楠，王驰. 基于偏振光谱融合的镜面反射去除方法及其在植被健康监测中的应用[J]. 光谱学与光谱分析, 2023, 43(11): 3607-3614.
LI Si-yuan, JIAO Jian-nan, WANG Chi. Specular Reflection Removal Method Based on Polarization Spectrum Fusion and Its Application in Vegetation Health Monitoring. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2023, 43(11): 3607-3614.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2023)11-3607-08 或 https://www.gpxygpfx.com/CN/Y2023/V43/I11/3607

[1] Thomas S C. Global Change Biology, 2005, 11(2): 197.
[2] Sandmeier S R. Remote Sensing of Environment, 2000, 73(3): 257.
[3] XU Jin, MENG Ji-hua（徐晋，蒙继华）. Remote Sensing Technology and Application（遥感技术与应用），2016，31(1)：74.
[4] Lichtenthaler H K, Wellburn A R. Biochemical Society Transactions, 1983, 11(5): 591.
[5] Madeira A C, Mendonca A, Ferreira M E, et al. Communications in Soil Science and Plant Analysis, 2000, 31(5-6): 631.
[6] SU Wei, ZHAO Xiao-feng, SUN Zhong-ping, et al(苏伟，赵晓凤，孙中平). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2019, 39(5): 1535.
[7] Baranoski G V G, Rokne J G. International Journal of Remote Sensing, 2005, 26(3)：503.
[8] Kim S R, Lee W K, Lim C H, et al. Forests, 2018, 9(3): 115.
[9] Zang Z, Wang G X, Lin H, et al. ISPRS Journal of Photogrammetry and Remote Sensing, 2021, 171: 253.
[10] Sims D A, Gamon J A. Remote Sensing of Environment, 2002, 81(2-3): 337.
[11] Li Y, Chen Y, Huang J. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(4): 2173.
[12] Wang S K, Yu C Y, Sun Y J, et al. Multimedia Tools and Applications, 2018, 77(9): 11363.
[13] LIN Yi, LIU Si-yuan, YAN Lei, et al（林沂，刘思远，晏磊，等）. Spectroscopy and Spectral Analysis（光谱学与光谱分析），2020，40(6)：1956.
[14] SU Wei, XIE Zi-xuan, WANG Wei, et al（苏伟，谢茈萱，王伟，等）. Transactions of the Chinese Society for Agricultural Machinery（农业机械学报），2020，51(5)：173.
[15] Li S Y, Jiao J N, Wang C. Remote Sensing, 2021, 13(17): 3510.
[16] Lavigne D A, Breton M, Fournier G, et al. Proceedings of SPIE—The International Society for Optical Engineering, Orlando, FL, 13 April, 2009, 733673361A-73361A-11.
[17] Jordan C F. Ecology, 1969, 50(4): 663.
[18] Goldstein D H. SPIE International Symposium on Optical Science and Technology, San, Diego, 30 July, 2000, 4133: 112.
[19] Woolley J T. Plant Physiology，1971, 47(5): 656.
[20] Grant L. Remote Sensing of Environment, 1987, 22(2): 309.
[21] Bousquet L, Lachérade S, Jacquemoud S, et al. Remote Sensing of Environment, 2005, 98(2-3): 201.
[22] Rondeaux G, Vanderbilt V C. International Journal of Remote Sensing, 1993, 14(9): 1815.
[23] Grant L, Daughtry C S T, Vanderbilt V C. Environmental and Experimental Botany, 1987, 27(2): 139.
[24] Grant L, Daughtry C S T, Vanderbilt V C. Physiologia Plantarum, 1993, 88(1): 1.