Multi-feature object identification of remote sensing image based on vague soft sets

Abstract

Multi-feature classification and image segmentation are two cores in object-oriented classification method of high resolution remote sensing images. Multi-feature object identification is an important part of multi-feature classification, which is identification for the image regions or the segmentation objects segmented by image segmentation under the guidance of a corresponding relationship between objects and features or combination. A method of multi-feature object identification was proposed based on vague soft sets. Firstly, the vague soft sets were formed by building the parameter sets according to spectral characteristics and object-oriented features of the segmentation objects. Secondly, according to general TOPSIS (the Technique for Order Preference by Similarity to Ideal Solution), a TOPSIS based on Vague soft sets for multi-feature object identification was proposed, which obtained a object identification result of the segmentation objects by using similarity measure of vague soft sets to sort attribution of the cover types for the segmentation objects. The experimental results show that the proposed method obtains a correct result of object identification and is feasible and effective.