Improvement of the Example-Regression-Based Super-Resolution Land Cover Mapping Algorithm

摘要: Super-resolution mapping (SRM) is a method for generating fine-resolution land cover map from coarse-resolution fraction images. Example-regression-based SRM algorithms can estimate with detailed spatial information by learning patterns available maps. Existing example-regression-based are sensitive to errors, and the results often include many linear artifacts speckles. To overcome these shortcomings, this study proposes an improved algorithm. The objective function of proposed algorithm comprises three terms. first term used minimize difference between values estimated input values. second maximize class membership possibility fine pixels in result. final make result locally smooth. compared several popular using both synthetic real Experimental indicate that produce less speckles artifacts, more details, smoother boundaries, higher accuracies than comparison.