|
Shadow extraction of building using fusion of edge and point feature orientation from high resolution satellite imagery
|
Farzane Yousefiyan *   , Hamin Ebadi , Amin Sedaghat |
| K.N.Toosi University of Technology |
|
|
Abstract: (2870 Views) |
| Shadow detection is an important preprocessing step in many applications of remote sensing, particularly in high resolution images. Shadow represents information about the shape, relative position and direction of the object and in urban environments occupy a significant portion of the image. Shadow can have positive and negative effects in objects interpretation. Shadows can be regarded as a type of useful information in building position recognition and height estimation. Researchers have presented model-based methods, property-based methods and based on index to shadow detection. In this research, we use local feature to identify the shadow of buildings in order to detect building position recognition. After point local feature extraction, we estimate their orientation. Orientation histogram calculate and due to the perpendicular edges of different buildings can extract main orientations. Edge map definition, which emphasizes edges only in the main orientations and dilate in the main orientations. This improved edge map is fused with initial shadow features. Edges are used to detect shadow of building and remove shadows that are not for the building. The proposed method was performed on four high-resolution image. Ground reference data obtained manually and Recall, precision and F-score, in best situation, respectively 90.1, 88.86 and 89.48 indicate the efficiency and proper performance of the proposed method. |
|
| Keywords: Main Orientation, Edge Detection, Shadow Detection, Fusion, Point Local Feature. |
|
|
Full-Text [PDF 1599 kb]
(807 Downloads)
|
Type of Study: Research |
Subject:
Aerial Photogrammetry
Received: 2018/12/25 | Accepted: 2018/12/25 | Published: 2018/12/25
|
|
|
|
|
|
