Automatic building extraction from very high-resolution image and LiDAR data with SVM algorithm

Abstract

Recent developments in the field of remote sensing have introduced new sensor technologies in usage of LiDAR, SAR, and high-resolution optical data. Classification performance is expected to increase through combining these various data sources. The purpose of this study is to develop a new approach for automatic extraction of buildings in urbanized and suburbanized areas. For this purpose, multi-feature extraction process including the spatial, spectral, and textural features were conducted on the very high spatial resolution multispectral aerial images and the LiDAR data set. SVM algorithm was trained by using this multi-feature data, and the classification was performed. After the classification of building and non-building, objects were extracted with high accuracy for the test areas. As a result, it has been proven that multi-features derived from combination of optical and LiDAR data can be successfully applied to solve the problem of automatic detection of buildings by using the proposed approach.