Kyuheon Kim

An effective method for combining multiple features of image retrieval

We propose an image retrieval system using the Borda count method in combining multiple experts (CME). It combines color-, shape- and texture-based retrieval systems. CME method can complementarily combine results of each retrieval system, which uses different features. There are some problems when the Borda count method in pattern recognition is applied to image retrieval. Thus, we propose a modified Borda count method to solve these problems. In the experiment our method reduces false positive errors and produces better results than that of each retrieval module that uses only one feature.

Interactive contents authoring system based on XMT and BIFS

As the Internet has been widely used for all the area of industries, the traditional characteristics of communication and broadcasting has been merged into a new service such as interactivity based broadcasting. Under an interactive environment, viewers not only watch broadcasting programs provided from a contents provider such as traditional broadcastings, but also pass their requirements to a service provider, obtain supplementary information and search broadcasting programs on the basis of their individual component objects. In order to access a program in terms of its component contents, thus, it is required to compose a scene and have reactions on the basis of its individual objects rather than a whole program itself. Therefore, this paper introduces a new authoring system, which can easily and conveniently produce an interactive contents-based broadcasting program by using MPEG-4 technologies. The authoring system presented in this paper describes a program in terms of two formats such as a textual format (XMT) for readability and a binary format (BiFS) for transmission. Since the XMT format is a XML-like one, it is easier for a user to understand and edit a scene composition. Alterantively, BiFS is suitable for transmission because of its binary property.

A study on path planning for mobile robot based on fuzzy logic controller

In this paper, we propose an efficient navigation algorithm employing fuzzy theory, which does not need analytical models on an environment. A technique that a single module can drive multiple ultrasonic sensors is presented. In addition, an ultrasonic sensor drive circuitry design for detection and avoidance of unknown obstacles is described. Finally, this paper presents the efficiency of the proposed fuzzy algorithm both through simulation and through floor test.

A 3DTV Broadcasting Scheme for High-Quality Stereoscopic Content Over a Hybrid Network

Various methods are used to provide three-dimensional television (3DTV) services over a terrestrial broadcasting network. However, these services cannot provide high-quality 3D content to consumers. It is because current terrestrial broadcast networks are allocated with limited bandwidths to transmit 3D content while 3D content requires larger bandwidth compared to that of 2D content. To overcome this limitation, this paper proposes a hybrid 3DTV broadcasting system, which utilizes both a terrestrial broadcast network and a broadband network. In the proposed system, two elementary streams of left and right views for a stereoscopic video service are transmitted over a terrestrial broadcasting network and a broadband network, respectively. In addition, the proposed system suggests a new mechanism for synchronization between these two elementary streams. The proposed scheme can provide high-quality 3DTV service regardless of bandwidth of the terrestrial broadcast network while maintaining backward compatibility with a 2D DTV broadcasting service.

Hybrid interactive data service architecture over T-DMB with mobile network

There have been lots of technical developments in broadcasting field since 1990s. Providing high quality video and audio is one improvement and supporting mobile reception in a high-speed vehicle is another. Terrestrial Digital Multimedia Broadcasting (T-DMB) is one of the successfully deployed technologies to provide mobile broadcasting service. One of the major breakthroughs achieved by T-DMB in addition to robust vehicular reception is a system architecture naturally enabling integrated interactive data services, which is defined by MPEG-4 System architecture composed of Object Descriptor Framework and Binary Format for Scene (BiFS) technology. However, interactive service in T-DMB has two fundamental limitations. First, interactive data service should be always rendered on top of a video due to the T-DMB specification. Second, initial data for interactive service is only received by broadcasting channel. These limitations prevent the implementation of emerging data service recently deployed widely by the smart handheld devices such as personalized data service customized for the user characteristics or location-based data service. Therefore, this paper proposes the architecture of hybrid interactive data service (HIDS) for T-DMB to overcome these limitations by defining a hybrid BIFS, which supports individual scene descriptions delivered from broadcasting and mobile channels, and is accessed from the local memory. This proposed architecture also maintains backward compatibility with the already deployed legacy T-DMB receivers.

Efficient stereoscopic contents file format on the basis of ISO base media file format

A lot of 3D contents haven been widely used for multimedia services, however, real 3D video contents have been adopted for a limited applications such as a specially designed 3D cinema. This is because of the difficulty of capturing real 3D video contents and the limitation of display devices available in a market. However, diverse types of display devices for stereoscopic video contents for real 3D video contents have been recently released in a market. Especially, a mobile phone with a stereoscopic camera has been released in a market, which provides a user as a consumer to have more realistic experiences without glasses, and also, as a content creator to take stereoscopic images or record the stereoscopic video contents. However, a user can only store and display these acquired stereoscopic contents with his/her own devices due to the non-existence of a common file format for these contents. This limitation causes a user not share his/her contents with any other users, which makes it difficult the relevant market to stereoscopic contents is getting expanded. Therefore, this paper proposes the common file format on the basis of ISO base media file format for stereoscopic contents, which enables users to store and exchange pure stereoscopic contents. This technology is also currently under development for an international standard of MPEG as being called as a stereoscopic video application format.

Development of an interactive contents authoring system for MPEG-4

The authoring system we present provides a powerful graphical user interface that helps a content creator produce MPEG-4 interactive contents easily and conveniently. In the presented authoring system, MPEG-4 content can be produced as two types of format such as a binary (MP4) and a textual format (XMT, extensible MPEG-4 textual). Especially, the presented authoring system manipulates arbitrarily shaped video objects and, also, provides a semi-automatic segmentation tool for extracting an arbitrarily shaped region of interest in a video object.

Design and implementation of MPEG-4 authoring tool

Under an interactive broadcasting environment, viewers have been requesting additional and linked information of individual contents in TV programs. In order to satisfy these demands from viewers, first of all, it is required to develop a new authoring tool that can edit AV (Audio Visual) contents as independent objects and add related information of the contents. In this paper, we propose an MPEG-4 based authoring tool. The characteristics of our authoring tool are that it generates an XMT file for an object and a scene description, which is then coded into an mp4 file according to the MPEG-4 specification. In order to show the usefulness of the proposed authoring tool, rote show the authored AV contents rising the reference software (IM1 2D player) provided by MPEG.

Peer-to-peer scalable video streaming using Raptor code

This paper proposes a method for saving resources of the video distribution server by using peer-to-peer communications. In order to adapt to a time-varying channel, scalable video coding and a fountain code, a Raptor code is used. The Raptor code protects video quality asymmetrically according to the significance of scalable video layers; varying in peer-to-peer networks, as a forward error correction (FEC) method. It is also a more efficient peer-to-peer distribution tool (because of its rateless feature) than conventional erasure recovery FEC such as the Reed Solomon code. The simulation shows that peer-to-peer data exchange between the clients can save the resources of the server.

The simple aggregation format for lightweight applications scene representation (LASeR)

The recent mobile multimedia services such as video download and flash like graphic interface are not enough to satisfy demands from end-users. LASeR is a standard to improve and make efficient rich-media applications and services on mobile devices. Using the standard, the advanced services can be provided, such as mobile TV service, Web-like navigation service, interactive location-based services, and so on. In this standard, a method for simple aggregation format (SAF) are provided, which is a streaming-ready format for packaging scenes and media together and streaming them onto such protocols as HTTP, TCP, MPEG-2, and so on. In this paper, we introduce LASeR and give a detailed technical overview of SAF. Also, we will show how different between MPEG-4 systems and LASeR