Effectivity of Prediction Enhancement of Land Cover Classification for Remote Sensing Images Using Automatic Feature Learning Model

Abstract

The proposed approach presents a cost-effective and environmentally friendly solution for classifying land use in urban areas. It relies on optical aerial imagery and decision trees generated from unmanned aircraft systems (UAS) to extract land cover information. The extracted data is then combined with a possession parcel map to establish a connection between land use and cover. The decision tree algorithm takes into account the geometric characteristics of parcels to create a prepared land use parcel map. This approach is versatile and can be applied to different scales of aerial imagery, making it well-suited for city planning and landscape monitoring applications. The technique employs object-oriented image analysis, and the analytic hierarchy process is used to determine the optimal scale for segmenting and classifying images. Image segmentation on various scales is utilized to identify the main land.