Monitoring road pavement deformations by ICP-based stereophotogrammetric method

Abstract

Introduction.  Regular road surface monitoring is essential for an objective assessment of road conditions and timely intervention to combat emerging deformations. Traditional survey methods have long been limited to visual inspection, while developing pavement microcracks could remain undetected for a long time, subsequently leading to significant repair costs. A pressing issue is the development of a road surface monitoring methodology that can detect changes in road surface geometry, including microrelief analysis. Methods and Materials. In modern practice, a detailed assessment of road surface conditions uses a three-dimensional representation of the surface as an array of laser reflection points obtained by terrestrial and mobile laser scanning. Road surface changes, including deformations, can be monitored and assessed by comparing point clouds obtained using stereophotogrammetry at different times. This requires minimizing the influence of external orientation errors on the clouds obtained at different times by aligning them with each other. A prototype equipped with a full-frame digital camera and a GNSS geodetic receiver was used. Between surveys conducted along two routes approximately one hour apart, changes in the asphalt surface during the specified time were simulated using specialized monitoring mockups to ensure that the survey data reflected both actual pavement deformations and simulated changes. The resulting point clouds contained over 40 million points each. To optimize the analysis, they were thinned to a density of 5 mm and cropped at the boundaries of the study area. The coordinates of the image projection centers were determined in RTK mode. The thinned point clouds were imported into Cloud Compare software (ENST, France) for analysis. Results.  For the study, photogrammetric models of the surveys were constructed, with point clouds superimposed on one another using exterior orientation parameters. The analysis revealed deviations that made it impossible to align the point cloud superposition results. Using the ICP algorithm, a reanalysis of the point clouds was performed after mutual orientation, shifting the second cloud relative to the first using a stable contour. Models placed after the initial survey showed actual changes. It was established that the accuracy of cloud construction based on the  original exterior orientation parameters of the images was insufficient, and after integrating the point clouds using ICP algorithms, display improvements were evident. Discussion and Conclusion.  Photogrammetric survey systems exist that are relatively simple to use and relatively inexpensive, as exemplified by the presented model. It has been established that using only the coordinates of projection centers obtained using GNSS measurements as the initial data will result in insufficient point cloud orientation accuracy, specifically for deformation monitoring. However, the system is simple in design, and using modern ICP algorithms, it is possible to align two different time-lapse images, achieving millimeter-level accuracy in deformation changes. These images can be used to detect dynamic road cracks and predict road surface changes. This system, as a relatively inexpensive and simple piece of equipment, can be used for municipal needs, as well as for assessing deformations in specific areas (such as utility punctures or projectile impacts on road sections), to facilitate rapid response.