Road Extraction in a very high-resolution image based on Hough transformation and LBP operator

Abstract

This paper presents a novel algorithm of road detection in an image with very high spatial resolution based on Local binary patterns and Hough transformation. The very high resolution also allows a real representation of roads on a map, but also causes a significant increase in noise. This article proposes a road detection method using only the digital image as a source of information. It can detect and determine both the orientation of the road by exploiting some of their intrinsic properties by using the Hough transformation. This transformation dedicated to the extraction of the lines; for his power to extract linear structure. This approach generally provides good results in spite of some disadvantage. In particular, only the linear parts of the roads will be extracted, and in case of non-linearity method fails to extract these parts of roads. That is the reason why we will add texture presented by the LBP operators that enhance detection by removing the false detections found by Hough detector. The LBP method, introduced by Ojala is defined as a measure of invariant texture, derived from a texture in a local neighborhood.

The proposed approach has been tested on different images. The provided results demonstrate the effectiveness of the proposed method.