Spectral volume index creation and performance evaluation: A preliminary test for tree species identification

To fully mine information regarding differences among various tree species from remote sensing data and improve the accuracy of recognition, this study utilized spectral reflection value, wavelength, time as parameters employed three algorithms to create an expression for volume index (SVI). Then, were obtained by applying RedEdge-MX four phases, SVI features extracted, a mixed feature set band + texture digital surface model was constructed. A random forest algorithm determine importance derive optimal classification. The additional objectives if have active role in classification which is more conducive extracting useful features. extracted with constraints exhibit better performance recognition than those without constraints. Moreover, using variable-constrained constant-constrained volume. combination could (the highest overall 92.76%), but improvement effect limited value 92.16% when not combined). These findings show that method be used difference images certain extent hence, identification.