Segmentation of surface elevation points derived from high-resolution UAV imagery is used to identify terrain and non-terrain points. Grey level thresholding of radiance values associated with terrain points enables extraction of impervious surfaces like Roads and Platforms. A morphometric based algorithm along with threshold length to width ratio is used to extract more precisely the roads out of other ground features. The experimental results show that this method can detect better linear features. Some comparatively small mask and boundaries are detected with changes in the local statistics by the automatic methods. Further, the discontinuity is detected and eliminated with morphometric parameters to minimize the effects of interfering objects and shadow effects. The extracted road network is evaluated for both spatial and semantic accuracy by comparing with the existing road network. This method demonstrated better extraction of linear features like roads than conventional image classification methods.
Automatic Extraction of Roads from UAV Using Thresholding and Morphometric Parameters
Lecture Notes in Civil Engineering
International Conference on Unmanned Aerial System in Geomatics ; 2019 ; Roorkee, India April 06, 2019 - April 07, 2019
23.02.2020
7 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Automatic extraction of main roads using aerial images
British Library Online Contents | 2010
|Face Recognition using Binary Thresholding for Features Extraction
British Library Conference Proceedings | 1999
|Edges: saliency measures and automatic thresholding
British Library Online Contents | 1997
|