A Fractal Evaluation Method Of Pavement Cracks Based On Homography Solution System With On-vehicle Active Vision

With the rapid development of the transportation industry,China has become one of the busiest countries in the aspect of road transportation task in the world.In order to extend the service life of highways,reduce the occurrence of traffic accidents and improve the social economic benefits of transportation,it is necessary to maintain the highway regularly and timely.Acquiring road surface information quickly and accurately is the most important task for highway maintenance management.Pavement crack is a common road surface distress.The size and shape of pavement cracks affect the capacity of transportation tasks of road surface directly.In addition,traditional evaluation methods of pavement crack lack quantitative description of the pavement-crack shape in researching on a fractal evaluation method of pavement cracks based on homography solution system with on-vehicle active vision is of great significance to acquire the road surface information quickly and accurately.A homography solution system with on-vehicle active vision is designed for the sake of reconstructing the real scale pavement cracks,improving the accuracy of pavement-crack skeleton reconstruction,and enhancing the convenience of on-vehicle pavement-crack detection system,aiming at the current research status of the pavement cracks on-vehicle detection method.A real-scale pavement cracks skeleton reconstruction method which based on active vision are studied,and solution of pavement-crack image homography matrix and the extraction of pavement-crack skeleton are completed.In order to reduce the influence of uneven illumination intensity on the extraction of pavement-crack skeleton,a pavement-crack image segmentation method based on fractional Brown motion is studied.A fractal evaluation method of pavement cracks which based on box-dimension is studied for classifying the similar length but different shapes pavement cracks.The homography solution system with on-vehicle active vision utilizes four linestructured lights which are relatively fixed to the camera that is used to collect the images of the pavement.Firstly,the line-structure lights are calibrated for obtaining the coordinates of the line-structure lights in the camera coordinate system.Secondly,according to the coordinate of the line-structure lights in the camera coordinate system and the internal parameters of the camera,the coordinate of the projection points on the road surface in the camera coordinate system are calculated.Finally,pavement-crack image homography matrix can be corrected by the coordinates of projection points which are produced by the intersections of the line-structured lights and the pavement.The coordinates are in the camera coordinate system and the image coordinate system.In order to improve the effect of pavement-crack skeleton extracted from pavement crack images,bilateral filtering and logarithmic transformation enhancement are applied to pavement crack images.The extraction of the pavement-crack skeleton is realized through Open CV.The reconstruction of the real pavement cracks is completed by using the homography matrix of the pavement image.In order to extract the pavement crack skeletons more thoroughly,the pavement crack images are segmented and reconstructed based on the window size and Brown fractal dimension.Using the box dimension to evaluate pavement-crack skeletons with real scale information quantitatively.According to the dimension of the pavement-crack skeleton,the pavement cracks could be classified.