Multi-spectrum data supervised classification method based on relevant linear information entropy

Abstract

The invention discloses a multi-spectrum data supervised classification method based on relevant linear information entropy and relates to a remote sensing image supervised classification method based on relevant linear information entropy and a kNN classifier. The supervised classification method comprises the following steps of: firstly, automatically filtrating training sample sets required for supervised classification after manually sampling; secondly, automatically optimizing the parameters for determining the kNN classifier algorithm; thirdly, classifying multi-spectral remote-sensing images by using the kNN classifier algorithm. The supervised classification method is characterized in that the classification precision of the kNN classifier algorithm can be effectively improved and the classification time is reduced by efficiently and automatically filtrating training samples input by multiple variables and automatically optimizing the classifier parameters, so that the supervised classification method is more appropriate for the high-precision classification task of the multi-spectral remote-sensing images with large data volume.

Description

technical field [0001] The invention belongs to the field of remote sensing data interpretation, and relates to a supervised classification method for remote sensing images based on linear correlation information entropy and a kNN classifier. Background technique [0002] At present, most remote sensing imaging systems in application worldwide are multi-spectral, they can simultaneously acquire images in several bands, and provide multiple snapshot images of spectral characteristics, which are more valuable than single-band images, in the field of classification applications It can also achieve better classification confidence than single-band images. [0003] The topics of classification are different in different applications. For supervised classification (or supervised classification for short), the analyst must select representative pixels for each category as the training area. If the multispectral images contain extremely rich and unique visual cues, methods such as ...

AI Technical Summary

Problems solved by technology

Although the kNN algorithm also relies on the limit theorem in principle, it is only related to a very small number of adjacent samples when making category decisions.

Images