The Stereo Vision Method Application To Road Inspection

The paper presents the stereo vision method for the mapping of road pavement. The mapped road is a set of points in three-dimensional space. The proposed method of measurement and its implementation make it possible to generate a precise mapping of a road surface with a resolution of 1 mm in transverse, longitudinal and vertical dimensions. Such accurate mapping of the road is the effect of application of stereo images based on image processing technologies. The use of matching measure CoVar, at the stage of images matching, help eliminate corner detection and filter stereo images, maintaining the effectiveness of the algorithm mapping. The proper analysis of image-based data and application of mathematical transformations enable to determine many types of distresses such as potholes, patches, bleedings, cracks, ruts and roughness. The paper also aims at comparing the results of proposed solution and reference test-bench. The statistical analysis of the differences permits the judgment of error types.