Hyperspectral imaging with a band matrix reduction method to detect early drought stress in tomato

Drought stress is one of the key abiotic stresses affecting plant growth, crop yield and food quality. The main objective this study to investigate potential effectiveness hyperspectral imaging with band selection method for rapid detection early drought tomatoes. First, unsupervised algorithm - K-means statistical histogram are used extract samples representing each experimental treatment group. Then, solve problems related high redundancy correlation data, matrix reduction (BMRM) based on recursive feature elimination theory proposed determine optimal subset. constructed according ranking obtained by discrimination coefficient -Coefi, which calculated from average spectral curve first-derivative spectrum. Finally, waveband algorithms was validated comparison successive projections algorithm, competitive adaptive reweighted sampling, cross-validation full results demonstrated that BMRM achieved higher classification accuracy fewer bands selected, amount calculation not greatly improved. provides a more accurate, effective way detecting stress.