Open AccessNon-destructive determination of Malondialdehyde (MDA) distribution in oilseed rape leaves by laboratory scale NIR hyperspectral imaging
Author(s) -
Wenwen Kong,
Fei Liu,
Chu Zhang,
Jianfeng Zhang,
Hailin Feng
Publication year - 2016
Publication title -
scientific reports
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.24
H-Index - 213
ISSN - 2045-2322
DOI - 10.1038/srep35393
Subject(s) - hyperspectral imaging , malondialdehyde , calibration , partial least squares regression , preprocessor , remote sensing , sampling (signal processing) , environmental science , computer science , artificial intelligence , biological system , mathematics , chemistry , biology , machine learning , statistics , computer vision , biochemistry , filter (signal processing) , antioxidant , geology
The feasibility of hyperspectral imaging with 400–1000 nm was investigated to detect malondialdehyde (MDA) content in oilseed rape leaves under herbicide stress. After comparing the performance of different preprocessing methods, linear and nonlinear calibration models, the optimal prediction performance was achieved by extreme learning machine (ELM) model with only 23 wavelengths selected by competitive adaptive reweighted sampling (CARS), and the result was R P = 0.929 and RMSEP = 2.951. Furthermore, MDA distribution map was successfully achieved by partial least squares (PLS) model with CARS. This study indicated that hyperspectral imaging technology provided a fast and nondestructive solution for MDA content detection in plant leaves.