TVICS: An Efficient Traffic Video Information Converting System Hang Yue Mid-America Transportation Center Civil Engineering Department University of Nebraska-Lincoln Lincoln, NE 68588, USA (until April 30 th , 2012) Email: yuehang366@gmail.com Peter Z. Revesz Computer Science & Engineering Department University of Nebraska-Lincoln Lincoln, NE 68588, USA Email: revesz@unl.cse.edu Abstract — This paper presents a new system called TVICS that converts traffic video data into vehicular motion information in spatio-temporal databases. The TVICS system interpolates the vehicular trajectory data (time, location and velocity), which are extracted from vi deo, and integrates them with spatial road information for the storage of dynamic transportation environments. TVICS can avoi d the storage issues caused by traffic videos with their associated large data size. Moreover, users can manage and operate multiform and multidimensional traffic data in a spatio-temporal transportation environment. Experimental results show that TVICS has a high accuracy in transportation applications. Keywords – GIS; traffic video; spatio-temporal database; vehicular velocity interpolation; transportation I. INT RODUCT ION In this paper we consider transportation applications. Some transportation-related applications, such as urban planning, require only static spatial databases or geographic information systems (GISs). For instance, Miller [1] used a GIS for the evaluation of traffic analysis zone (TAZ) effects, the design of optimal zoning systems, and the derivation of better zonal distance measures. In addition, various intelligent transportation systems (ITS) often use static GIS map databases for location referencing and frequently exchange spatial information with other map databases [2]. However, more interesting transportation applications need to consider the values of traffic parameters that vary continuously over time. Spatial database systems deal with these data sets in an inefficient way via discrete time points or intervals. Given traffic data’s multiform and multidimensional nature, more efficient traffic data archiving is needed to add a temporal dimension to GIS-based transportation management systems. Today, video cameras are widely used for traffic monitoring and data collection. The combination of space and time is a defining feature of digital video [3]. However, considering the large space and expensive cost in traffic video storage, traffic video data are usually saved into video segments, scenes, shots, or frames [4]. Moreover, in current video management systems the storage manner of traffic videos is defined on the physical level, which does not allow the expression of high-level spatio-temporal relationships among traffic data [5, 6]. Therefore, it is difficult for current video database systems to quickly scan traffic video data and find the desired transportation- related spatio-temporal query results. We have looked at alternatives for storing video data. First, we can recognize that once the video cameras collect discrete vehicular trajectory data (some number of frames per second), and the trajectory data can be interpolated into continuous traffic data. Data model can describe the movement of vehicles with some functions of a temporal parameter t and spatial parameters. Second, such continuous traffic data can be conveniently stored in various types of spatio-temporal databases. For example, constraint databases can describe continuous spatio- temporal data in arbitrarily high-dimension [7]. At the same time, constraint databases allow various high-level query languages, including SQL and Datalog and their extensions [7]. The aim of this paper is to describe the development and features of an efficient Traffic Video Information Converting System (TVICS). Our TVICS system can convert traffic video data into spatio-temporal transportation databases. The system allows the users to choose various data interpolation options. In addition, we integrate into the system not only standard SQL queries but also high-level queries that are specifically designed for transportation-related applications. II. OVERVIEW OF THE TVICS SYSTEM The overall design and data management plan of the TVICS system are illustrated in Fig. 1. The TVICS system can optimize traffic data completeness and offer high-level spatio-temporal queries of transportation data. The design and development of the TVICS system consists of the following four main parts:  Data extraction: by the video-capture methods [8, 11], traffic video data extraction provides vehicular trajectory data (i.e. vehicular instantaneous location, time, and speed data points).