基于图像提取技术计算夏玉米覆盖度和反演叶面积指数的精度评价

doi:10.3969/j.issn.1000-6362.2016.04.013

摘要/Abstract

摘要：

如何将农作物从复杂动态变化的农田场景中准确提取出来，是作物分割方法后续提取覆盖度或反演叶面积指数准确与否的关键。本文以郑州、泰安和固城站2011年和2012年生长季的夏玉米为研究对象，利用在线式图像自动传输装置实时获取户外不同光照强度以及真实农田复杂背景下的夏玉米群体动态图像，在对原始图像进行几何校正的基础上，采用AP-HI、ExG、ExGR和CIVE4种作物分割方法提取夏玉米覆盖度和反演叶面积指数，通过对比试验定量评价每种作物分割方法对农田复杂背景的适应能力和精度，并从中选取适合农田复杂环境下夏玉米冠层图像覆盖度和叶面积指数的有效提取方法。结果表明：光照强度变化以及植物阴影、植物残渣等复杂背景对作物分割算法的精确性影响较大，AP-HI方法在光照适应性和对抗复杂环境两个方面均优于其它方法，相对误差在0.2以下，高于目估法测量的精度；通过对比分析发现，利用4种作物分割方法通过冠层孔隙率估算不同生长期夏玉米LAI，AP-HI反演模型最优，其估算的LAI与实际测得的LAI值比较的相关系数最高，为0.89~0.96,均方根误差最小，为0.47~0.75。综合考虑模型的精度和稳定性认为，基于AP-HI方法反演叶面积指数的方法具有一定应用优势。

关键词: 农作物, 图像处理, 覆盖度, LAI, 模型

Abstract:

How to extract accurately the crops from the complex field scene is the key to calculate crop coverage and invert LAI for crop segmentation methods. In this paper, the dynamic images of summer maize growth season in 2011-2012, in Zhengzhou,Taian and Gucheng, which were under the different light intensity and complex outdoor background with shadows, plants residues, were obtained through on-line, real-time automatic transmission device. Meanwhile, to overcome the error caused by image distortion, geometric correction of the raw images was needed, crop coverage extraction ability and leaf area index inversion performance of four popular crop extraction algorithms (ExG, ExGR, CIVE and AP-HI) were compared and evaluated. By comparison, the effective extraction method for canopy coverage and leaf area index of summer maize under complex environment was selected. On this basis, models of canopy coverage and leaf area index with canopy porosity method were established and verified with measured data. The results showed that the light intensity changes and complex field environment, which contained plant shadows and residues, had a significant impact on the accuracy of crop segmentation algorithm. And inversion model of AP-HI was superior to the other methods in both light adaptability and complex environment, the relative error compared with true image was less than 0.2 and higher than the current visual estimation accuracy. Leaf area index was estimated by four extraction algorithms in summer maize growth season in 2011 and 2012. Based on comparison of R and RMSR among models, high fitting models were selected. The optimal model for LAI was based on AP-HI extraction algorithm, which had the highest R (0.89-0.96) and the lowest RMSR (0.47-0.75). Considering the accuracy and stability of the model, inversion model of AP-HI based on the method of application had more advantages.

Key words: Crops, Image processing, Coverage, Leaf area index, Model

李翠娜，张雪芬，余正泓，王秀芳. 基于图像提取技术计算夏玉米覆盖度和反演叶面积指数的精度评价[J]. 中国农业气象, 2016, 37(04): 479-491.

LI Cui-na, ZHANG Xue-fen, YU Zheng-hong, WANG Xiu-fang. Accuracy Evaluation of Summer Maize Coverage and Leaf Area Index Inversion Based on Images Extraction Technology[J]. Chinese Journal of Agrometeorology, 2016, 37(04): 479-491.

参考文献

[1]瞿瑛,刘素红,谢云,等.植被覆盖度计算机模拟模型与参数敏感性分析[J].作物学报,2008,34(11):1964-1969. Qu Y,Liu S H,Xie Y,et al.Computer simulation model of fractional vegetation cover and its parameters sensitivity[J].Acta Agronomica Sinic,2008,34(11):1964-1969.(in Chinese) [2]张学艺,郭建茂,韩颖娟,等.基于植被指数的宁夏灌区春小麦叶面积指数模型[J].中国农业气象,2011,32(2):279-282. Zhang X Y,Guo J M,Han Y J,et al.LAI model of spring wheat in Ningxia irrigated area based on MODIS-VI[J].Chinese Journal of Agrometeorology,2011,32(2):279-282.(in Chinese) [3]Weiss M,Troufleau D,Baret F,et al.Coupling canopy functioning and radiative transfer models for remote sensing data assimilation[J].Agricultural and Forest Meteorology, 2001,108(2):113-128. [4]秦伟,朱清科,张学霞,等.植被覆盖度及其测算方法研究进展[J].西北农林科技大学学报(自然科学版),2006,34(9): 163-168. Qin W,Zhu Q K,Zhang X X,et al.Review of vegetation covering and its measuring and calculating method[J].Jour. of Northwest Sci-Tech Univ. of Agri. and For.(Nat. Sci. Ed.),2006,34(9):163-168.(in Chinese) [5]Zhou Q,Robson M.Automated range land vegetation cover and density estimation using ground digital images and a spectral-contextual classifier[J].Remote Sensing,2001,22(17): 3457-3470. [6]章文波,符素华,刘宝元,等.目估法测量植被覆盖度的精度分析[J].北京师范大学学报(自然科学版),2001,37(3):402-407. Zhang W B,Fu S H,Liu B Y,et al.Error assessment of visual estimation plant coverage[J].Journal of Beijing Normal University(Natural Science),2001,37(3):402-407.(in Chinese) [7]王桂琴,郑丽敏,朱虹,等.图像处理技术在冬小麦叶面积指数测定中的应用[J].麦类作物学报,2004, 24(4):108-112. Wang G Q,Zheng L M,Zhu H,et al.Application of image processing technology in wheat canopy leaf area index measuring[J].Journal of Triticeae Crops,2004,24(4):108-112. (in Chinese) [8]李荣春,陶洪斌,张竹琴,等.基于图像处理技术的夏玉米长势监测研究[J].玉米科学,2010,18(2):128-132. Li R C,Tao H B,Zhang Z Q,et al.Study on summer maize group growth monitoring based on image processing technique[J].Journal of Maize Sciences,2010,18(2):128-132. (in Chinese) [9]李存军,王纪华,刘良云,等.基于数字照片特征的小麦覆盖度自动提取研究[J].浙江大学学报(农业与生命科学版), 2004,30(6) :650-656. Li C J,Wang J H,Liu L Y,et al.Automated digital image analyses for estimating percent ground cover of winter wheat based on object features[J].Journal of Zhejiang University (Agriculture & Life Sciences),2004,30(6):650-656.(in Chinese) [10]陆明,申双和,王春艳,等.基于图像识别技术的夏玉米生育期识别方法初探[J].中国农业气象,2011,32(3):423-429. Lu M,Shen S H,Wang C Y,et al.Initial exploration of maize phonological stage based on image recognition[J].Chinese Journal of Agrometeorology,2011,32(3):423-429.(in Chinese) [11]任杰,柏延臣,王锦地,等.从数码照片中快速提取植被覆盖度的方法研究[J].遥感技术与应用,2010,25(5):719-723. Ren J,Bai Y C,Wang J D,et al.An efficient method for extracting vegetation coverage from digital photographs[J]. Remote Sensing Technology and Application,2010,25(5): 719-723.(in Chinese) [12]陆秀明,黄庆,孙雪晨,等.图像处理技术估测水稻叶面积指数的研究[J].中国农学通报,2011,27(3):65-68 Lu X M,Huang Q,Sun X C,et al.Application of image processing technology in rice canopy leaf area index[J].Chinese Agricultural Science Bulletin,2011,27(3): 65-68.(in Chinese) [13]张雪芬,薛红喜,孙涵,等.自动农业气象观测系统功能与设计[J].应用气象学报,2012,23(1):105-112. Zhang X F,Xue H X,Sun H,et al.Function and designing of automatic observing system for agro-meteorology[J].Journal of Applied Meteorological Science,2012,23(1):105-112.(in Chinese) [14]Sakamoto T,Shibayama M,Kimura A,et al.Assessment of digital camera-derived vegetation indices in quantitative monitoring of seasonal rice growth[J].ISPRS Journal of Photogrammetry and Remote Sensing,2011,66(6):872-882. [15]Shibayama M,Sakamoto T,Takada E,et al.Regression-based models to predict rice leaf area index using biennial fixed point continuous observations of near infrared digital Images[J].Plant Production Science,2011,14(4):365-376. [16]赵继强,谢东辉,孟夏,等.规则林地覆盖度照相测量法的精度分析[J].农业工程学报,2013,29(19):152-158. Zhao J Q,Xie D H,Meng X.Accuracy analysis of vegetation cover measurement using digital photography for regular planted woodland[J].Transactions of the CSAE,2013,29(19): 152-158.(in Chinese) [17]孙涛,刘振波,葛云健,等.基于数码相片Gamma校正的水稻叶面积指数估算[J].生态学报,2014,37(13):3548-3557. Sun T,Liu Z B,Ge Y J,et al.Estimation of paddy rice leaf area index based on photo gamma correction[J].Acta Ecologica Sinica,2014,37(13):3548-3557.(in Chinese) [18]中国气象局.农业气象观测规范[M].北京:气象出版社, 1993. China Meteorological Administration.Specification for agricultural meteorological observation[M].Beijing :China Meteorological Press,1993.(in Chinese) [19]Zhang Z.Flexible camera calibration by viewing a plane from unknown orientations[A]. Proceedings of the Seventh IEEE International Conference on Computer Vision[C]. USA,1999:666-673. [20]Yu Z,Cao Z,Wu X,et al.Automatic image-based detection technology for two critical growth stages of maize: emergence and three-leaf stage[J].Agricultural and Forest Meteorology,2013,174-175:65-84. [21]Frey B J,Dueck D.Clustering by passing messages between data points[J].Science,2007,315(5814): 972-976. [22]Woebbecke D M,Meyer G E,von Bargen K,et al.Color indices for weed identification under various soil,residue and lighting conditions[J].Transaction of the American Society of Agricultural Engineering,1995,38(1):259-269. [23]Neto J C.A combined statistical-soft computing approach for classification and mapping weed species in minimum-tillage systems[D].University of Nebraska,NE,2004. [24]Kataoka T,Kaneko T,Okamoto H,et al.Crop growth estimation system using machine vision[A].The 2003 IEEE/ASME International Conference on Advanced Intelligent Mechatronics[C]. USA,2003 [25]Otsu N.A threshold selection method from gray-level histogram[J].IEEE Transactions on System Man and Cybernetics,1979,11(285-296): 23-27. [26]Nilson T.A theoretical analysis of the frequency of gaps in plant sands[J].Agric Meteorol,1971,8:25-38. [27]陈江鲁,王克如,李少昆,等.基于宽范围动态植被指数的棉花冠层覆盖度监测[J].棉花学报,2011,23(3):265-271. Chen J L,Wang K R,Li S K,et al.Monitoring of the cotton vegetation fraction based on wide dynamic range vegetation index[J].Cotton Science,2011,23(3):265-271.(in Chinese) [28]Ewing R P,Horton R.Quantitative color image analysis of agronomic images[J].Agron J,1999,91:148-153.