Hyperspectral Visualization of Citrus Leaf Moisture Content Based on CARS-CNN
DAI Qiu-fang1, 2, 3, LIAO Chen-long1, 2, LI Zhen1, 2, 3*，SONG Shu-ran1, 2, 3，XUE Xiu-yun1, 2, 3，XIONG Shi-lu1, 2
1. College of Electronic Engineering (College of Artificial Intelligence), South China Agricultural University，Guangzhou 510642, China
2. Division of Citrus Machinery, China Agriculture Research System, Guangzhou 510642, China
3. Information Monitoring Engineering Technology Research Center, Guangzhou 510642, China
Abstract：Water deficit of citrus leaves is one of the important factors affecting the growth of citrus. In order to study the effect of water stress on the moisture content of citrus, hyperspectral technology was used to rapidly and non-destructively detect the moisture content of citrus leaves, and pseudo-color processing was applied to realize the visualization of moisture content. 100 citrus leaves were collected, and 500 leaves with different gradient moisture content were obtained by drying method. The samples were divided into a training set (350 samples) and a testing set (150 samples) according to the ratio of 7∶3. The determination coefficient (R2) and root mean square error (RMSE) was used to evaluate the model’s prediction quality. A convolution neural network (CNN) is used to predict spectrum data. The CNN model uses a one-dimensional convolution kernel with three convolution pooling layers activated by the RELU activation function. The output layer uses a linear activation function for regression prediction, and the nadam algorithm is used to optimize and update the model with 1 000 epochs; The Raw spectrum data and the spectrum data are pretreated by SG, MSC and SNV are used respectively. The full bands, the feature bands screened by CARS and the feature bands extracted by PCA are imported into the CNN model respectively. The best model is CARS-CNN of the Raw spectrum data, the R2c and RMSEC of the training set are 0.967 9 and 0.016 3 respectively. The R2v and RMSEV of the testing set are 0.947 0 and 0.021 4, respectively. The effect of the full bands CNN model of the Raw spectrum data is the second, and the R2c and RMSEC of the training set are 0.934 3 and 0.024 9, respectively. The R2v and RMSEV of the testing set are 0.915 9 and 0.028 6, respectively; At the same time, the best combined results of the support vector machine regression model (SVR), partial least squares regression model (PLSR) and random forest model (RF) under different pretreatment methods and characteristic wavelength selection were compared. The best combination model (Raw spectrum+CARS+PLSR, SNV+PCA+RF, SNV+PCA+SVR) was compared with CARS-CNN of Raw spectrum data, CARS-CNN model still has the best prediction effect. Compared with other models, the CARS-CNN model has higher prediction accuracy than SVR, PLSR and RF models, after further feature extraction by CARS and convolution kernel. Select the trained CARS-CNN model, import the hyperspectral image into the model, calculate the moisture content of each pixel, and get the pseudo-color image, which can more intuitively display the visual distribution of leaf moisture content. The result provides a faster, more intuitive and more comprehensive assessment of citrus leaf moisture content, a basis for the study of citrus leaf water stress, and a reference for optimising intelligent irrigation decision-making.
[1] CHEN Fei，LI Hong-ping，CUI Ning-bo（陈 飞，李鸿平，崔宁博）. Agricultural Research in the Arid Areas（干旱地区农业研究），2021，39(3)：42.
[2] LI Hong-ping，CUI Ning-bo，CHEN Yu-xin（李鸿平，崔宁博，陈昱辛）. Journal of Irrigation and Drainage（灌溉排水学报）, 2019，38(10)：1.
[3] ZHANG Xiao-xing，FAN Yi，JIA Yue，et al（张效星，樊 毅，贾 悦，等）. Transactions of the Chinese Society of Agricultural Engineering（农业工程学报），2018，34(3)：143.
[4] Panigrahi P，Srivastava A K. Scientia Horticulturae，2016，210：6.
[5] Murphy R J, Whelan B, Chlingaryan A，et al. Precision Agriculture，2018，20：767.
[6] Zhu Zhen，Li Tiansheng，Cui Jing, et al. Acta Agriculturae Scandinavica, Section B—Soil & Plant Science, 2020, doi:10.1080/09064710.2020.1726999.
[7] Malek Salim，Melgani Farid，Bazi Yakoub. Journal of Chemometrics，2018，32(5): e2977.
[8] Dong Jialin，Hong Mingjian, Xu Yi, et al. Journal of Chemometrics，2019，33: e3184.
[9] Acquarelli J, Van Laarhoven T, Gerretzen J，et al. Analytica Chimica Acta, 2017，954：22.
[10] ZHONG Liang，GUO Xi，GUO Jia-xin，et al(钟 亮, 郭 熙, 国佳欣，等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报)，2021，37(1)：203.
[11] Jie Dengfei，Wu Shuang, Wang Ping, et al. Food Analytical Methods，2021，14：280.
[12] Song Xiangzhong，Du Guorong, Li Qianqian, et al. Analytical and Bioanalytical Chemistry，2020，412: 2795.
[13] Kumar Keshav. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy，2021，244: 118874.
[14] Guerra-Urzola R，Van Deun K, Vera J C, et al. Psychometrika，2021, 86(4): 893.
