Identification of Aphid Infection on Rape Pods Using Hyperspectral Imaging Combined with Image Processing
YU Hao1, Lü Mei-qiao2, LIU Li-ming3, YU Gui-ping4, ZHAO Yan-ru5, HE Yong5*
1. Ningbo Institute of Technology, Zhejiang University, Ningbo 315100, China
2. Jinhua Polytechnic, Jinhua 321017, China
3. Zhejiang Technical Institute of Economics, Hangzhou 310018, China
4. Zhejiang Academy of Agricultural Machinery, Jinhua 321017, China
5. College of Biosystems Engineering and Food Science, Zhejiang University, Hangzhou 310058, China
Abstract:Rape aphids can reduce the production and quality of rapeseed seriously, so early discrimination of the rape aphids and identification of the infection location are helpful for precisely spraying pesticide. In this study, hyperspectral imaging in visible and near-infrared region combined with imaging processing were employed to discriminate the healthy and aphid infected rape pods, as well as identify the location of rape aphids. Here, a total of 323 samples covering 138 healthy and 185 aphid-infected rape pods was investigated. Firstly, principal component analysis (PCA) was used to conduct the cluster analysis of the two groups rape pods, and the wavelength at 737 nm selected by X-loading was considered as an important waveband for the purpose of aphid discrimination. Then, statistical analysis of spectral data from the two groups’ samples at single band (737 nm) was finished by boxplot. At the same time, a linear equation y=2.917 6-3.345 7x (x represented the spectral data of 737 nm, y denoted the predicted dummy classes) was obtained based on above analysis. Relying on the linear equation, discriminant analysis was carried out for the 323 samples and the recognition accuracy reached 99.0%. Next, the location of rape pods was identified based on the single band grayscale images. For the infected rape pods, the method led to an overall detection accuracy of 81.1%. The results revealed that the spectral data at 737 nm and its image information is a promising tool for identifying the location of aphids in rape pods, which could provide a theoretical reference and basis for designing the handheld detection system and the precise spraying of rape industry in the further work.
[1] FU Ming-lian, HOU Shu-min, ZHU Xuan, et al(符明联, 侯树敏, 朱 炫, 等). Journal of Anhui Agriculture Sciences(安徽农业科学), 2011, 39(9): 5287.
[2] Bock C H, Poole G H, Parker P E, et al. Critical Reviews in Plant Sciences, 2010, 29(2): 59.
[3] Rumpf T, Mahlein A K, Steiner U, et al. Computers and Electronics in Agriculture, 2010, 74(1): 91.
[4] Mirik M, Michels Jr G J, Kassymzhanova-Mirik S, et al. Computers and Electronics in Agriculture, 2007, 57(2): 123.
[5] LUO Ju-hua, HUANG Mu-yi, ZHAO Jin-ling, et al(罗菊花, 黄木易, 赵晋陵, 等). Transactions of the Chinese Society of Agricultural Engineering (农业工程学报), 2011, 27(7): 215.
[6] Prabhakar M, Prasad Y G, Thirupathi M, et al. Computer and Electronics in Agriculture, 2011, 79(2): 189.
[7] Zhao Y, He Y, Xu N. Computer and Electronics in Agriculture, 2012, 89: 41.
[8] Zhao Y R, Yu K Q, Li X L, et al. Scientific Reports, 2016, 6: 38878.
[9] Li X L, Nie P C, Qiu Z J, et al. Expert Systems with Applications, 2011, 38(9): 11149.
[10] Yu K Q, Zhao Y R, Liu Z Y, et al. Food and Bioprocess Technology, 2014, 7(11): 3077.
[11] Yu K Q, Zhao Y R, Li X L, et al. Plos One, 2014, 9(12): e116205.
[12] Filella I,Penuelas J. International Journal of Remote Sensing, 1994, 15(7): 1459.
[13] Gitelson A A, Merzlyak M N, Lichtenthaler H K. Journal of Plant Physiology, 1996, 148(3): 501.