Towards automatic road extraction from digital images

Abstract EN

Automatic extraction of road networks from digital images has been an active research subject for over a decade.

This is due to its significant role in various application areas, like Intelligent Transportation (IT) Systems as well as GIS applications.

While the manual extraction of road data is time consuming, the urgent necessity is automatic extraction of road data.

In this paper, an original framework for automatic extraction of road networks from digital images is developed.

The Proposed algorithm is based on the theories of image processing that is oriented towards the mathematical morphological operation that is based on the theories set for spatial structure analysis.

Implementing the road extraction process passed through two main stages : the first was extracting street network from a raster image and defining the road centerline and the second was identifying the coordinates of intersection points of the streets network.

In order to guarantee and validate the methodology of the proposal a software package was developed.

Moreover, a case study is conducted to analyze the efficiency and performance of the proposed algorithms.