A Method for Automatic Road Extraction of High Resolution SAR Imagery
- PDF / 1,907,882 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 23 Downloads / 185 Views
RESEARCH ARTICLE
A Method for Automatic Road Extraction of High Resolution SAR Imagery M. Saati & J. Amini & M. Maboudi
Received: 13 November 2014 / Accepted: 4 February 2015 # Indian Society of Remote Sensing 2015
Abstract Nowadays automatic road extraction from satellite imageries is considered as one of the most important research trends in the field of remote sensing. This paper presents a method for automatic extraction of road centerlines from synthetic aperture radar (SAR) imagery. During the first step, three features, namely the direction of the least total radiance, the corresponding radiance, and the contrast are extracted to define the road characteristics by the backscatter coefficient of each pixel and its neighboring pixels from the SAR imagery. The fusion of the extracted features is carried out in the next step for detection of the road areas by using a fuzzy inference system. Afterwards, the morphology skeletonization is applied on the road areas to extract the road skeleton. Then some interested seed points are extracted so that they could be used in a snake model, which is employed to connect the seed points in order to form up the road centerlines. The proposed algorithm is tested on different parts of TerraSAR-X images. The experimental results reveal that the proposed method is effective in terms of correctness, completeness, and quality.
Keywords Road extraction . Fuzzy algorithm . High resolution . Synthetic aperture radar . Active contour model
M. Saati : J. Amini (*) Department of Surveying and Geomatics Engineering, College of Engineering, University of Tehran, Kargar-e- Shomali St., Tehran, Iran e-mail: [email protected] M. Saati e-mail: [email protected] M. Maboudi Islamic Azad University, Qazvin Branch, Qazvin, Iran
Introduction Manual extraction of the objects through the satellite imagery done by expert operators is claimed to be really costly and time consuming. Therefore, automatic extraction of the objects from the images is considered as a fundamental research area in the context of remote sensing for mapping applications. Of those objects of interest, roads have been the most frequently appeared objects in the establishment of maps in urban and sub urban areas (Zarrinpanjeh et al. 2013). When disaster strikes, roads play pivotal roles in bringing relief provisions to the disaster-struck areas. Subsequently, road obstruction information would be necessary for the prompt delivery of the aids. Unfortunately, it is very common that obtaining adequate optical images would be difficult due to bad weather conditions (e.g., cloud coverage) and lack of day/night acquisition capability. Several synthetic aperture radar (SAR) sensors can instead provide wide spatial coverage of the earth, as their sensing capability stays intact throughout day/night and in almost any weather condition. Thus, road extraction from SAR images is complementary or even alternative to optical remote sensing images. Currently, automatic road extraction from high resolution (HR) SAR data is a high profile resea
Data Loading...
