A self-adaptive selective method of remote sensing image classification algorithms

Xin Pan, Hongbin Sun

COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(3) 98-103

School of Computer Project & Technology, Changchun Institute of Technology, Changchun City, Jilin Province, China, 130012

Remote sensing image classification algorithms, which can obtain information of land usecover quickly and inexpensively have been widely used in the field of GIS. The quality of classification results is not only affected by the quality of remote sensing data, but also affected by the character of classification algorithm. At present, despite a lot of algorithms have been proposed, but users usually meet difficulties in algorithm selection due to single classification algorithm cat not applicable to all classification cases. This study proposes a self-adaptive selective method for remote sensing image classification algorithms based on data complexity evaluation, through data complexity evaluation, our method can distinguish remote sensing data’s character even from same satellite sensor and give user recommendation of algorithm selection. Experiments indicate that the algorithms selected by this method can achieve higher classification accuracy, which provides the recommendation for the selection of appropriate classification models to users.