THREE-DIMENTIONAL ROAD SURFACE ESTIMATION USING A MONOCULAR CAMERA INSTALLED IN A VEHICLE

Abstract

 In road infrastructure maintenance, accurately assessing and managing pavement conditions is crucial. However, existing methods face challenges in terms of efficiency, precision, and cost, highlighting the need for high-efficiency, high-precision, and cost-effective techniques. This study proposes an advanced method for evaluating wheel ruts using a smartphone-mounted camera and image analysis. The proposed approach enhances accuracy through feature-matching-based bird’s-eye view correction and template matching to improve distortion evaluation. Additionally, it extends to 3D road surface shape estimation, including longitudinal local road conditions such as potholes and unevenness at the joint. Experimental validation confirms that the method can effectively assess various road surface features in three dimensions.