Research by Associate Professor Hwanjo Yu (CSE) and Dr. Youngwoo Kim (CSE) was accepted to Knowledge-Based Systems, a world class SCI journal. The research was written when Dr. Kim was just a sophomore.
They developed a new efficient indexing method, called “GeoTree”. GeoTree was designed to be used with Geoferenced Video, which incorporates spatiotemporal information such as time, location and camera direction captured by cameras, smartphones or car black boxes. Minimum Bounding Rectangle (MBR), the traditional method, is not suitable for describing location and direction of moving scenes continuously changing because it describes an area in spatial, not substantial information. To overcome the problem, they adopted a new data structure, Minimum Bounding Tilted Rectangle (MBTR). It stores meaningful parts of moving scenes so it reduces the index size substantially and improves the search speed for Georeferenced Video data. It will support searching for videos with adjustable viewable angles and distances.