Estimation of Eggplant Leaf Nitrogen Content Based on Hyperspectral Imaging and Convolutional Auto-Encoders Networks
WANG Hao-yu1, 2, 3, WEI Zi-yuan1, 2, 3, YANG Yong-xia1, 2, 3, HOU Jun-ying1, 2, 3, SUN Zhang-tong1, 2, 3, HU Jin1, 2, 3*
1. College of Mechanical and Electronic Engineering, Northwest A&F University, Yangling 712100, China
2. Key Laboratory of Agricultural Internet of Things, Ministry of Agriculture and Rural Affairs, Yangling 712100, China
3. Key Laboratory of Agricultural Information Awareness and Intelligent Services, Yangling 712100, China
Abstract Leaf nitrogen content (LNC) is crucial for assessing plant growth status and photosynthetic capacity. Accurate LNC can aid in the rational control of nitrogen fertilizer application, which is critical for achieving efficient agricultural production. Chemical analysis methods can accurately detect nitrogen content. However, it often requires destructive sampling and cumbersome steps, which are difficult to measure in real-time. Spectral technology can enable nondestructive detection of LNC, but the high dimensionality and noise inherent in spectral data make accurate estimation challenging for precision agriculture. To achieve accurate prediction of nitrogen content in eggplant leaves, this paper proposed a feature extraction method of spectral data based on hyperspectral imaging (HSI) technology and a one-dimensional convolutional autoencoder network (CAE). The proposed method utilized pixel-level spectral data to train the CAE, fully utilizing the HSI data of leaves. This can extract deep features that retain local spectral features related to the nitrogen content distribution on the leaf surface, reducing data dimension, filtering out noise, and enhancing the accuracy and stability of the nitrogen content prediction model. In this paper, we set up four nitrogen application gradients for eggplant, obtained leaf samples with varying nitrogen content using culture, and measured their HIS data. Multiple scattering correction algorithm was used for data preprocessing. The HSI-CAE method, competitive adaptive reweighting (CARS) algorithm, and random frog (RF) algorithm were used to extract spectral data's deep features and characteristic wavelengths, respectively. The partial least squares regression(PLSR)models were built based on these features. The influence of deep features and different feature wavelength combinations on the accuracy of the prediction model was compared to determine the optimal feature extraction method. The results were as follows: the test set determination coefficient of the prediction model, established by using deep features from the CAE encoder of different depths, was greater than 0.85. When 28-dimensional features were output, the test set determination coefficient was 0.910 2, and the root mean square error was 3.118 9 mg·g-1. It was found that the CAE-PLSR model has the best prediction performance, which verified the feasibility and superiority of the HSI-CAE feature extraction method for estimating nitrogen content in eggplant leaves. In conclusion, the HSI-CAE feature extraction method can efficiently analyze HSI data and extract its deep features. These features contain information highly related to nitrogen content. The deep feature modeling used in this research greatly reduced the complexity of the model. It effectively improved the accuracy of the nitrogen content prediction model, providing a new way of implementing accurate prediction of nitrogen content based on the HSI technology.
Corresponding Authors:
HU Jin
E-mail: hujin007@nwafu.edu.cn
Cite this article:
WANG Hao-yu,WEI Zi-yuan,YANG Yong-xia, et al. Estimation of Eggplant Leaf Nitrogen Content Based on Hyperspectral Imaging and Convolutional Auto-Encoders Networks[J]. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2024, 44(08): 2208-2215.
[1] Raj R, Walker J P, Pingale R, et al. International Journal of Applied Earth Observation and Geoinformation, 2021, 104: 102584.
[2] Mwinuka P R, Mourice S K, Mbungu W B, et al. Agricultural Water Management, 2022, 266: 107516.
[3] Jiang J, Zhu J, Wang X, et al. Remote Sensing, 2021, 13(4): 739.
[4] SONG Xiang-zhong, TANG Guo, ZHANG Lu-da, et al(宋相中,唐 果,张录达, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2017, 37(4): 1048.
[5] Yu X, Lu H, Liu Q. Chemometrics and Intelligent Laboratory Systems, 2018, 172: 188.
[6] Wang C, Liu B, Liu L, et al. Artificial Intelligence Review, 2021, 54(7): 5205.
[7] Fu L, Sun J, Wang S, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2022, 281: 121641.
[8] Yang D, Yuan J, Chang Q, et al. Infrared Physics & Technology, 2020, 109: 103412.
[9] YUAN Fei-niu, ZHANG Lin, SHI Jin-ting, et al(袁非牛,章 琳,史劲亭, 等). Chinese Journal of Computers(计算机学报), 2019, 42(1): 203.
[10] Pang L, Men S, Yan L, et al. IEEE Access, 2020, 8: 123026.
[11] Miao X, Miao Y, Liu Y, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2023, 284: 121733.
[12] Li X, Wei Z, Peng F, et al. Computers and Electronics in Agriculture, 2022, 198: 107036.
[13] Li D, Li C, Yao Y, et al. Computers and Electronics in Agriculture, 2020, 174: 105459.
[14] Yang L, Deng S, Ma S, et al. Computers & Electrical Engineering, 2022, 98: 107648.
[15] Sanaeifar A, Yang C, de la Guardia M, et al. Science of the Total Environment, 2023, 861: 160652.