Multimedia database development for a highway information system

Photographic logging systems used by highway agencies provide engineers with information in the analysis of traffic accidents, design improvements, and highway pavement management. However, there exist limitations for such systems in the areas of accessibility, search capability of the image library, and synchronization of video data with traditional engineering site data. The capability of offering multiple user accesses to highway video footages cannot be provided by current systems. The first part of the dissertation discusses the development of a new system called the multimedia-based highway information system (MMHIS). MMHIS uses digital video technology, high-speed networking, and video server to solve the problems in existing systems. The digital video is compressed using motion-JPEG or MPEG method. A new database system is designed to store engineering site data in a more organized way. Site data retrieval is more efficient than old systems. A synchronization algorithm is also designed to synchronize the site data display with the full-motion highway video. MMHIS can also handle turning movements when the video reaches an intersection or an exit ramp. In addition, a 3-D visualization method is used in MMHIS to present a 3-D map to the user. The system has the ability to zoom and pan the map. Location information about any point on the map can be queried. Highway information can also be obtained if the queried point is on a road. Engineers can get access to the databases through computer networks without leaving their offices.; MMHIS can present pavement surface images and the corresponding distress data to the user. Currently the main method of collecting pavement surface distress data is through manual inspection. The second part of the dissertation summarizes the development in the field of automatic pavement surface distress survey. Existing working systems and their limitations are presented in this part. Image processing methods including neural network methods needed in the automatic pavement surface distress survey are also presented.