Building detection using local features and DSM data

Abstract

Detecting and locating buildings in satellite images has various application areas. Unfortunately, manually detecting buildings is hard and very time consuming. Therefore, in the literature several methods are proposed to automatically detect buildings. These methods can be divided into two main groups. In the first group, researchers used panchromatic or multispectral information to detect buildings. In the second group, researchers used DSM data to detect buildings. In this study, we propose two novel methods to detect buildings by combining the panchromatic and DSM data. The first method uses corner points extracted by Harris corner detection method. These corner points are used jointly with DSM data. Using a kernel based density estimation method, possible building locations are detected. In the second method, shadow of buildings are used in a similar way. We tested both methods on WorldView-2 images and DSM data generated from them.

Ground filtering and DTM generation from DSM data using probabilistic voting and segmentation

“…The first step in this operation is extracting seed points. We utilized DSM data to detect buildings with a probabilistic voting method (Özcan, Ünsalan, and Reinartz 2013) in a previous study. We extend this method further to detect seed points from DMS data in this study.…”

Building change detection with RGB-D map generated from UAV images

“…On the base of stereo imagery and DSM, Tian et al [1] proposed a change detection method by the joint use of height changes and Kullback-Leibler divergence similarity measure between the original images. Two methods were proposed by Ozcan et al [2] to detect building change by combining the panchromatic and DSM data. And both methods were tested on WorldView-2 images and DSM data generated from them.…”

“…There are many applications which make use of 3D geometry models and radiometric images from satellites or special aircrafts. A straightforward way to detect changed areas is comparing two 3D geometry models captured at different moments [1][2][3]. However, producing an accurate 3D geometry model is costly which usually involves high resolution aerial stereo rig or LIDAR mounted on the manned aerial vehicle followed by some complex registration procedures [4].…”

“…With the generated RGB-D map, our framework can handle large illumination change while keeping the whole procedure easily to be operated compared with the generation of accurate 3D geometry model. (2) A coarse-to-fine point cloud registration method is proposed. We use the RGB map for coarse registration , then the feature point clouds and the original images are used for fine registration.…”

A probabilistic approach to building change detection

“…We specifically focused on building change detection, since after a natural disaster damaged building information is one of the most important one. Our building change detection method is based on our previous study focusing on probabilistic building detection [10]. For more detail on building detection methods, readers can also check our previous study [11].…”