Classification of Tree Species in Different Seasons and Regions Based on Leaf Hyperspectral Images

This paper aims to establish a tree species identification model suitable for different seasons and regions based on leaf hyperspectral images, mine more effective algorithm. Firstly, the reflectance spectra of leaves in were analyzed. Then, solve problem that 0-element sparse random (SR) coding matrices affects classification performance error-correcting output codes (ECOC), two versions supervision-mechanism-based ECOC algorithms, namely SM-ECOC-V1 SM-ECOC-V2, proposed this paper. In addition, algorithms was compared with six traditional all bands feature bands. The experiment results show seasonal regional changes have an effect leaves, especially near-infrared region 760–1000 nm. When spectral information is added into model, can be effectively classified. SM-ECOC-V2 achieves best both Furthermore, outperform method under SR strategy, indicating methods avoid influence matrix performance.