Post-processing of interpolation-based super-resolution mapping with morphological filtering and fraction refilling

Abstract

Interpolation-based super-resolution mapping (SRM) is a popular model to produce a super-resolution land-cover map from coarse-resolution fraction images. This model can maintain the holistic land-cover features; however, it also results in a super-resolution land-cover map that includes many speckle and linear artefacts, due to errors caused by both the interpolation and the label assignment steps. In this article, we propose a novel two-step post-processing algorithm for interpolation-based SRM. The first step is morphological filtering, which is used to eliminate artefacts and to preserve land-cover features in the super-resolution land-cover map produced by interpolation-based SRM. The second step is fraction refilling, which is applied to make the fraction constraints satisfied and the super-resolution land-cover map locally smooth. Based on the application to three simulated images with various interpolation algorithms and morphological filter operations, the performance of the proposed post-processing algorithm was assessed. The results show that the proposed post-processing algorithm increases the accuracy of the super-resolution land-cover map and is suitable for different interpolation-based SRM models.