Subpixel land cover mapping based on a new spatial attraction model with spatial-spectral information

Affiliation of Author(s):电子信息工程学院

Journal:INTERNATIONAL JOURNAL OF REMOTE SENSING

Key Words:HOPFIELD NEURAL-NETWORK PIXEL

Abstract:Subpixel mapping (SPM) is a technique for handling mixed pixels to derive hard classification map at finer spatial resolution. Due to simplicity and explicit physical meanings, spatial attraction model (SAM) has been proven to be an effective SPM method. SAM methods mostly differ in the way in which they compute the spatial attraction, for example, subpixel/pixel spatial attraction model (SPSAM), subpixel/subpixel spatial attraction model (MSPSAM), and the mixed spatial attraction model (MSAM). However, these methods are difficult to fully utilize the spatial-spectral information of the original remote sensing image due to diversity of the land cover classes and the limitation of the resolution of the sensor. To solve this problem, spatial attraction model with spatial-spectral information (SAM-SS) is proposed here. SAM-SS can pick up more spatial-spectral information of the original image by adding a new processing path, namely interpolation followed by spectral unmixing. Based on visual comparison and quantitative accuracy assessment, the SAM-SS shows the best performance in the four experimental results.

ISSN No.:0143-1161

Translation or Not:no

Date of Publication:2019-08-18

Co-author:wyq,Leung, Henry

Correspondence Author:王鹏

Date of Publication:2019-08-18