An Enhanced Rule-Based Fuzzy Segmentation Approach for Automated Urban Feature Extraction Using High-Resolution Satellite Imagery

Abstract

A fuzzy segmentation approach based on rule-based fuzzy rules is proposed in this study to obtain urban features using high-resolution satellite images. Multiresolution segmentation and spectral difference segmentation are combined to effectively identify and classify houses, roads, trees, and agricultural fields in urban areas and rural farms. A fuzzy rule set was developed using satellite datasets from IKONOS, LISS IV, and WorldView-2, improving classification accuracy. In this work, buildings were extracted from IKONOS images, agricultural fields were extracted from LISS IV images, and roads and vegetation were extracted from WorldView-2 images. The map updating capability was demonstrated for 1:2500 and 1:1000 scales, respectively, for buildings and agricultural fields. Furthermore, the gray-level cooccurrence matrix was employed to enhance classification reliability and mitigate spectral confusion. By automating the process, the need for additional GIS data is reduced, making it a cost-effective, scalable, and efficient approach. Compared to traditional manual feature extraction methods, this method is an effective alternative in urban planning, land use mapping, and environmental monitoring.