Water Content Detection of Potato Leaves Based on Hyperspectral Image. Issue 17 (2018)

• Record Type: Journal Article
• Authors: Sun, Hong; Liu, Ning; Wu, Li; Chen, Longsheng; Yang, Liwei; Li, Minzan; Zhang, Qin
• Is Part Of: IFAC-PapersOnLine. Volume 51:Issue 17(2018)
• Page Start: 443
• Page End: 448
• Publication Date: 2018
• Subjects: Water content -- potato leaves -- hyperspectral -- correlation analysis -- CARS; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2018.08.179
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: In order to indicate potato crop water content and guide precision irrigation, non-destructive water content detection of potato crop leaves was studied. Firstly, the spectral reflectance of 355 samples was collected by hyperspectral camera and the leaves water content was measured by weighing method. Secondly, the average reflectance of the whole leaves was extracted, and the sensitive wavelengths of leaf water content were screened respectively by correlation analysis (CA) and competitive adaptive reweighted sampling (CARS). The results were as follows: the 15 sensitive wavelengths located in the range of 1400-1450 nm were selected by CA method. While, there were 13 sensitive wavelengths selected by the CARS algorithm including 976.4 nm, 1037.7 nm, 1044.5 nm, 1061.4 nm, 1108.7 nm, 1139 nm, 1357.8 nm, 1380.7 nm, 1397 nm, 1432.8 nm, 1452.3 nm, 1513.6 nm and 1520.0 nm. Finally, after compared the partial least squares regression (PLSR) modeling results of the water content detection based on two group sensitive wavelengths. The CARS-PLSR was elected to detect the water content of potato leaves. The modeling calibration accuracy of CARS-PLSR was 0.9878, and the validation accuracy coefficient was 0.9366. It provides a new theoretical method for detecting water content of potato plant in the field.