3D Extraction of the Relief of Road Surface through Image Analysis
Majdi Khoudeir, Benjamin Bringier · 2016
From the point of view of image processing, road surfaces can be considered as textured images including both micro- and macro-textures. The extraction of relief is based on modeling the photometric behavior of the surface. As the Lambertian model is no longer valid and the relief extract is unreliable, one solution is the development of photometric models according to specular phenomena. However, these models are difficult to use in terms of inversion of equations. To overcome this, this chapter describes an original method which involves four steps. The first step determines the different image areas, specular and diffuse ones. In the second, the surface is taken as a set of micro-facet and their slopes are determined from the geometrical configuration of the image shooting system. The third step determines the slopes of micro-facets in diffuse areas from the model of Lambert. The last step reconstructs the three-dimensional relief of the entire surface from these slopes.