Hyperspectral Dimensionality Reduction of Forest Types Based on Cat Swarm Algorithm

One of the main ways of dimensionality reduction of hyperspectral image was band selection. The paper pro- posed a hyperspectral image bands selection method based on binary cat swarm algorithm to solve problems of the high complexity and intensive computation efficiently for follow-up applied research. In this paper, Jilin Wangqing Forestry Bureau was chosen as the study area, by optimization process of the cats' location electing, less associated and more in- formative bands were selected from 115 bands of HJ-1A, band combination (22,37,109), to distinguish 5 kinds of domi- nant tree species and get better classification accuracy.