| With the rapid development of highway construction, pavement inspection and maintenance management have become an important task for China highway construction. Cracks are one of the most important parameters to evaluate the surface quality and the early manifestations of most disease, which will affect the lifespan of the pavement and traffic safety directly. It can avoid serious problems caused by the crack further development if we detect cracks early and maintain them in time. Although scholars has been committed to research pavement crack inspection technology and data processing method, it’s still lacks practical and effective automatic detection scheme.The main purpose of this project is to study key issues of pavement crack inspection and focus on three key technologies for improving the accuracy of crack detection including pavement data acquisition, pavement cracks identification and cracks extraction. pavement information acquisition is the basis of analysis of pavement crack, so simple, rapid and comprehensive data acquisition technology is the key for improving the accuracy and automation level of pavement crack inspection. Due to pavement condition is complex, there are many interference factors such as texture, noise and shadows, it is inevitable to generate false positives and false negative error during2D and3D inspection. Based on the acquired data, how to identify the presence of cracks accurately and improving crack recognition rate are the primary tasks. Pavement crack extraction is the premise and base of accurate description and quantitative evaluation to pavement crack geometry features. When cracks exist, how to eliminate interference factors and extract cracks automaticly and accuratly is difficult in pavement inspection area. This study focus on the shortages of pavement data acquisition, pavement cracks identification and cracks extraction to conduct research and propose appropriate solutions.The main research works in this paper are as follows:Based on the detailed analysis of current research about pavement crack inspection technologies and the principal of2D image measurement and structure light3D technologies, for the problem of current inspection technologies, we propose a new pavement crack inspection technologies combining2D image and3D depth information and design collection devices and data processing program.For the problem of false positives and false negative error generating during the process of pavement cracks identification, we propose a pavement cracks identification method based on the fusion of2D image and3D depth information. In this proposed method,2D gray-scale image and3D laser scanning data are modeled as a mass function based on Dempster-Shafer (D-S) theory. For evaluating and validating the performance of the proposed method, a further discussion about the selection of thresholds and parameters of this model are given in details.The experimental results show that the proposed method takes advantage of the respective merits of2D images and3D laser scanning data and therefore improves the pavement crack detection accuracy and reduces recognition error rate compared to2D image intensity-based methods.Pavement condition is complex. There are many interference factors such as tire marks, oil spills, shadows, or repairs from the real pavement cracks. For this problem and low level of automation, we propose an automatic tracking and extraction method based on the fusion of2D image and3D depth information. In this proposed method, firstly, crack points extracted by3D structured light tracking are as seed points, then the enhanced image by singular operator was used to construct cost function and minimized energy diagram was calculated by fast marching algorithm, finally, the crack was extracted by the minimal path search technology. The experimental results show that this method can extract the crack accurately and improve accuracy and recall without human intervention, which can achieve automatic crack tracking extraction. |
PDF Full Text Request