Design And Realization Of Video Detection System Of Road Quality

Transportation business in our country has a rapid development. The quality of road is an important factor of transportation. The management and maintenance of the road has been a research focus on traffic engineering. The traditional Pavement Management System has some shortcomings, for example the acquisition of data is a hard work, data processing is complex, data is of low visualization, evaluation results are not intuitionistic etc. In this thesis, a video detection system of road quality is implemented to overcome above problems. The system is implemented on the platform of VC6.0. MapObjects of GIS redevelop component and ActiveX control are exploited to show road information by the way of graphics, video and decision text. The spatial data, attribute data and video data are managed synthetically and efficiently by Access database and ADO technology. Video retrieval and image processing are used to analysis and detect the pavement video effectively.This system is comprised of geographic information management module, road video management module and pavement fault detection module. The first module implements the functions of mapping, displaying control, layer control, statistical charts and road information querying. The second module mainly implements the management of video data and road attribute data, querying and detection of road video. The third module is the center of system. Automatic detection of fault road and querying of specific image are implemented in this module. In automatic detection, the traditional shot boundary detection and key-frame extraction have some defects on complexity and robustness. So this thesis proposes a new algorithm with adaptive threshold and interval detection. Simulations on this algorithm show that the efficiency of pavement fault video detection is improved effectively. Besides, a new method of selecting the key frame parameters is proposed which further improves the accuracy and scientific of automatic detection. Query and retrieval functions are implemented by the way of feature extraction and feature matching.Video detection system of road quality improves the science of the road quality management, ensures the road administration department to check road conditions visually and accurately, provides a favorable decision support for making conservation decision, has a strong practical value.