Region segmentation of multi-spectral remote sensing images using improved JSEG algorithm

Abstract

Image segmentation is the basic of extracting and recognizing interesting region,and it is also the key step of image analysis.Region-based JSEG algorithm is a new image segmentation algorithm considers not only color information but also spatial information,and also has a good effect in region segmentation for ordinary images and video images.In this paper,JSEG algorithm is introduced to segment remote sensing images,and is also improved to be applicable to segment multi-spectral remote sensing images and VHR remote sensing images which contain complicated texture characters.Experiments demonstrate that it can solve fuzzy region boundary problem effectively which is caused by complicated land object information.