Journal Article

A raster-based method for building simplification considering shape and texture features based on remote sensing images

Fan, Ruijie, Shen, Yilang

2025

Overview

Building simplification involves the process of reducing the complexity of building shapes and details, which is crucial for preserving key features, highlighting essential information, and enhancing map readability. As the map scale decreases, the complex details of buildings need to be effectively simplified. Although various methods exist for simplifying vector or raster data buildings, there is limited research on retaining the original building textures during the simplification process. This study proposes a raster-based method for the simplification and texturing of buildings in remote sensing imagery. The method begins with segmenting preprocessed individual raster data buildings using the superpixels extracted via energy-driven sampling (SEEDS) superpixel segmentation method. Superpixels to be retained are then selected based on the evaluation parameters, corner ratio (CR), and square ratio (SR). Subsequently, the building area texture is extracted, and suitable textures from the texture library are selected through texture feature comparison (TFC). The selected textures are then hue-adjusted to achieve simplified textures that closely resemble the original image. Compared to traditional raster-based building simplification methods, this superpixel-based approach for the simplification and texturing of buildings in remote sensing imagery provides more suitable textures for simplified structures. This enhances the effectiveness of raster-based map generalization, improving both the aesthetic appeal and functionality of maps.

• Publisher: Taylor & Francis Group
• Subject: Building simplification / map generalization / multi-scale representation / remote sensing imagery