La orientación de la textura como evidencia para la detección de caminos laterales en imágenes

Vision is one of the main sources of information for today driver assistance systems (DAS). Some DAS functionalities are lane departure warning, driver attention monitoring and automatic vehicle control, among others. Although visual perception systems have demonstrated great progress, some challenges still remain unresolved. One of these problems is the detection in an image of road topology changes as intersections, junctions and roundabouts. In this paper the behavior of the image texture orientation as an indicator of the existence of intersections with lateral roads is analyzed. The texture orientation is obtained by processing the image using Gabor wavelets filter banks. Experimental results with images captured in real environment and lighting conditions are presented to illustrate the proposed analysis.