Open AccessRoad Detection & Extraction for Autonomous vehicles Using Satellite Images
Author(s) -
Shelke Poonam Uttam,
Parul Arora,
Prof. D.A. Jadhav
Publication year - 2019
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.j9203.0881019
Subject(s) - computer science , pixel , satellite , track (disk drive) , ground truth , extraction (chemistry) , artificial intelligence , computer vision , transport engineering , engineering , chemistry , chromatography , aerospace engineering , operating system
Good roads play important role in positive growth of our civilization. They revolutionize and increase the life style of us and directly affect all our activities. Having record of up-to-date roads helps in taking quick decisions and pre-planning for various occasions like road trip or life critical situations like disasters. . It also helps in making maps for city planning and keeping track of expansion of city. Here, we have adopted morphological operations based techniques. All the images are resized to have same dimensions as they are from different sources. Comparison with ground truth images (manually extracted) is done to examine the performance of the algorithm. As this method is solely based on intensity of road pixels, the effect of different road structures and lane markings is absent. The algorithm has achieved 89% F1-score and 94% accuracy. Further improvement is required in algorithm to detect roads where the intensity of non-road parts (urban area, mowed agricultural land and other similar intensity structures) is similar to road pixels.