1. 中国农业大学现代精细农业系统集成研究教育部重点实验室，北京 100083
2. Center for Precision & Automated Agricultural System, Washington State University, Prosser, WA 99350, USA
Spectral Imaging Detection of Crop Chlorophyll Distribution Based on Optical Saturation Effect Correction
SUN Hong1, XING Zi-zheng1, QIAO Lang1, LONG Yao-wei1, GAO De-hua1, LI Min-zan1*, Qin Zhang2
1. Key Laboratory of Modern Precision Agriculture System Integration Research, China Agricultural University, Beijing 100083, China
2. Center for Precision & Automated Agricultural System, Washington State University, Prosser, WA 99350, USA
Abstract：Chlorophyll content is an important indicator of photosynthesis capability and nutrient content in crops. Measuring chlorophyll content of crops is considered to be the most effective method for detecting crop growth status. In this paper, a multi-spectral camera was built to capture images of maize plant in RGB(Red, Green, Blue) and NIR(Near Infrared) band, which was the fundamentalfor the distribution analysisof nutritional status with rapidand non-destructive method. The RGB and NIR images were acquired by image acquisition platform. Light saturation correction of RGB images based on light saturation correction algorithm was studied. The crop SPAD distribution map was established following the image matching and segmentation, image information extraction and correction. In the experiment, images of 15 maize plants were acquired by RGB-NIR camera, and 68 SPAD values were measured at different positions of the plants. Firstly, the RGB images were corrected by light saturation correction algorithm. At the same time, the NIR images were filtered and enhanced. Secondly, the RGB and NIR images were matched with SURF (Speeded-Up Robust Features) and RANSAC (Random Sample Consensus) algorithm. Used RGB images color feature, the mask was generated with ExG (Extra Green) and OTSU algorithm, and applied in the RGB -NIR images segmentation. The R, G, B and NIR components of the image were extracted and the reflectance were corrected by the fourth-order gray-scale plate. The Intensity(I) component histogram and crop SPAD value distribution were compared to verify the effect of the optical saturation correction algorithm. The results show that I component of RGB image concentrates between [140～180] before optical saturation correction, and between [85～130] after optical saturation correction because of correction in image blurring and RGB image saturation. The correlation coefficients between the image components (R, G, B, NIR) and gray scale reflectance were 0.829, 0.828, 0.745 and 0.994, respectively. Then the pseudo-color images of R, G, B and NIR bands were generated. The reflectance results (RNIR＞RG＞RR＞RB) indicated the spectral characteristics of crops which absorbed light in blue and red regions, reflected light in green and near infrared regions. Thirdly, the SPAD values at pixel level were calculated. The accuracy of chlorophyll content fitted in SPAD formula with R and NIR component reflectance before and after correction were compared. The R2 was 0.332 6 before correction and the R2 after correction was 0.619 3. Finally, the SPAD distribution map of crops was drawn, which could provide technical support for analyzing and monitoring the nutritional distribution of crops.
[1] YANG Zhang-qing, HU Jian-dong, DUAN Tie-cheng, et al(杨张青，胡建东，段铁城，等). Chinese Agricultural Science Bulletin(中国农业通报)，2019，35(7)：139.
[2] Zhang J, Han W, Huang L, et al. Sensors, 2016, 16(4): 437.
[3] SUN Hong, ZHAO Yi, ZHANG Meng, et al(孙 红, 赵 毅, 张 猛，等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2015, 31(S2): 186.
[4] LIU Zhen(刘 振). China Printing and Packaging Research(中国印刷与包装研究), 2013, 5(1): 23.
[5] Kober V I, Karnaukhov V N. Journal of Communications Technology & Electronics, 2016, 61(12): 1419.
[6] DING Yong-jun, LI Min-zan, SUN Hong, et al(丁永军, 李民赞, 孙 红，等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2012, 28(8): 175.
[7] Wang S, Zheng J, Hu H M, et al. IEEE Transactions on Image Processing, 2013, 22(9): 3538.
[8] HE Ming, WANG Ya-di, HE Jing(贺 明, 王亚弟, 贺 菁，等). Infrared and Laser Engineering(红外与激光工程), 2013, 42(10): 2620.
[9] XI Jia-qi, CHEN Xiao-dong, WANG Yi, et al(席佳祺, 陈晓冬, 汪 毅，等). Photoelectric Engineering(光电工程), 2019, 46(4): 47.
[10] ZHOU Fei, JIA Zhen-hong, YANG Jie, et al(周 飞, 贾振红, 杨 杰，等). Optoelectronics Laser(光电子激光), 2017, 28(5): 566.
[11] LIU Yang, ZHANG Jie, ZHANG Hui(刘 洋, 张 杰, 张 慧). Computerscience(计算机科学), 2018, 45(S1): 242.
[12] Nair D, Sankaran P. Journal of Visual Communication & Image Representation, 2018, 50: 9.
[13] Zhang L, Wang S, Wang X. IET Image Processing, 2018, 12(6): 1049.
[14] TAN Xiang, MAO Hai-ying, ZHI Xiao-dong, et al(谭 翔, 毛海颖, 支晓栋，等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2018, 38(2): 413.
[15] SUN Guo-xiang, WANG Xiao-yu, YAN Ting-ting, et al(孙国祥, 汪小旵, 闫婷婷，等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2014, 30(20): 187.
[16] Fu Longsheng, Elkamil Tola, Ahmad Al-Mallahi, et al. Biosystems Engineering, 2019, 183: 184.
