Sub-pixel mapping of remote sensing images based on sub-pixel/pixel spatial attraction models with anisotropic spatial dependence model

Abstract

Sub-pixel mapping (SPM) based on the Sub-pixel /Pixel Spatial Attraction Models (SPSAM) is a technique that improve the spatial resolution of remote sensing images. SPSAM-based SPM is based on the spatial dependence theory with isotropic assumption. In SPSAM, the weight calculation of spatial dependence is only relevant with the distance between the sub-pixel and its neighboring pixel. Obviously, the direction of spatial dependence is neglected. In this paper, a revised SPSAM-based SPM with anisotropic spatial dependence model (SPSAMA) is proposed. Sobel operator is utilized to determine the gradient magnitude and direction of coarse proportion image at every pixel. Then the gradient magnitude and direction will be used to reckon the weights of adjacent pixels proportions in the neighborhood. Experimental results demonstrated that the SPSAMA can produce land cover maps with greater accuracy than traditional SPSAM.