Evaluation of nitrogen content in cabbage seedlings using hyper-spectral images

Abstract

Monitoring of nutrient status of crops is essential for better management of crop production. Nitrogen is one of the most important elements in fertilizer for the growth and yield of vegetable crops. In this study, nitrogen content of cabbage seedlings was evaluated using hyper-spectral images. Cabbage seedlings, cultured at five nitrogen fertilization levels, were planted in the 128-cell plug trays and grown in a phytotron at National Taiwan University. The images, ranged from 410 to 1,090 nm, of cabbage seedlings were analyzed by a hyper-spectral imaging system consisting of CCD cameras with liquid crystal tunable filters (LCTF), which was developed in this study. The digital images of seedling canopies were processed including image segmentation, gray level calibration and absorbance conversion. Models including modified partial least square regression (MPLSR), step-wise multi-linear regression (SMLR) and artificial neural network with cross-learning strategy (ANN-CL) were developed for the determination of the nitrogen content in cabbage seedlings. The three significant wavelengths derived from SMLR model are 470, 710, and 1,080; and the best result is obtained by ANN-CL model, in which rc = 0.89, SEC = 6.41 mg/g, rv = 0.87, and SEV = 6.96 mg/g. The ANN-CL model is more suitable for the remote sensing in precision agriculture applications because not only its model accuracy but also only 3 wavelengths are needed.