Min-Ho Jeon

Network System Based on M2M for Efficient Visual Information Service

Current visual information is, as an information delivery mediator of information society, playing a very important role and its applied fields are growing exponentially. However, existing network system collecting and providing visual information requires human to act in awareness and transmission of data and causes data delay. In this paper, proposes network system replacing part processed by human awareness through combining M2M technology on the network system that collects visual information in order to solve the problem. The proposed system compares information from mobile devices captured on visual device and data for visual information requested by each server and sets the route visual information is transferred. As a result of test in environment to fit M2M major component with development of M2M devices using AVR, we could learn that only visual information matching content saved on M2M application server database built using transferred data from several servers among visual data collected continuously is transmitted to the server requesting information.

Location tracking system to improve the performance of video information service based on M2M network

A various range of services using image information has been provided in recent years. However, services specific for searching location status and providing people location information are insufficient. This study proposes a visual information service system that classifies images using information transmitted from a mobile object, and identifies a persons location and identity using screen coordinates and image processing algorithms. This system divides visual information into a cell unit by applying perspective projection formulas to image/video information collected by screen coordinates, and then organises blocks by substituting GPS data to each cell. Subsequently, the system senses human motion by comparing an image of current frame with previous frames, and recognises a person by comparing GPS data of motion detected area with real-time GPS data. According to test results, the recognition rate was lowered due to pixel noise generated by environmental factors. When tests were re-performed by eliminating external environmental parameters, it was found out that noises were minimised and motion detection was substantially improved.