• Journal of the Korea Society of Computer and Information
• /
• v.25 no.1
• /
• pp.55-61
• /
• 2020

With the development of deep learning technology and advances in computing power, video-based research is now gaining more and more attention. Video data contains a large amount of temporal and spatial information, which is the biggest difference compared with image data. It has a larger amount of data. It has attracted intense attention in computer vision. Among them, motion recognition is one of the research focuses. However, the action recognition of human in the video is extremely complex and challenging subject. Based on many research in human beings, we have found that artificial intelligence-like attention mechanisms are an efficient model for cognition. This efficient model is ideal for processing image information and complex continuous video information. We introduce this attention mechanism into video action recognition, paying attention to human actions in video and effectively improving recognition efficiency. In this paper, we propose a new 3D residual attention network using convolutional neural network based on two attention models to identify human action behavior in the video. An evaluation result of our model showed up to 90.7% accuracy.

• ETRI Journal
• /
• v.36 no.2
• /
• pp.232-241
• /
• 2014

A holographic display system with a 22-inch LCD panel is developed to provide a wide viewing angle and large holographic 3D image. It is realized by steering a narrow viewing window resulting from a very large pixel pitch compared to the wave length of the laser light. Point light sources and a lens array make it possible to arbitrarily control the position of the viewing window for a moving observer. The holographic display provides both eyes of the observer with a holographic 3D image using two vertically placed LCD panels and a beam splitter to support the holographic stereogram.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.153.2-153
• /
• 2001

A technique is described for producing a virtual object with realistic motion. A 3-D human motion model is obtained by applying a developed motion capturing technique to a real human in motion. Factorization method is a technique for recovering 3-D shape of a rigid object from a single video image stream without using camera parameters. The technique is extended for recovering 3-D human motions. The proposed system is composed of three fixed cameras which take video images of a human motion. Three obtained image sequences are analyzed to yield measurement matrices at individual sampling times, and they are merged into a single measurement matrix to which the factorization is applied and the 3-D human motion is recovered ...

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2018.07a
• /
• pp.116-118
• /
• 2018

본 논문에서는 기존의 동영상을 그랩컷(GrabCut) 알고리즘과 유니티3D를 이용하여 상호작용이 가능한 동영상을 제작하는 기법을 제안한다. 그랩컷 알고리즘을 이용하여 동영상에서 재생 프레임 단위로 원하는 객체 영역을 추출하고 흑백의 이미지로 이진화한다. 이진화된 결과물과 원본 동영상을 유니티3D에서 동시에 재생하면서 선택 영역의 이진화 픽셀 정보를 기반으로 사용자의 입력을 감지하는 동영상의 제작이 가능함을 보였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2014.11a
• /
• pp.214-215
• /
• 2014

3차원 비디오 부호화를 위한 표준안을 제정하기 위해 국제 표준화 기구인 JCT-3V(Joint Collaborative Team on 3D Video Coding Extension Development)에서는 3차원 비디오 부호화기술에 대한 표준화가 진행되고 있다. 본 논문은 현재 JCT-3V에서 HEVC(High Efficiency Video Coding) 기반으로 표준화가 진행 중인 3D-HEVC 부호화 기술들에 대해 살펴보고 그 부호화 및 복잡도 성능을 분석하였다. 이러한 성능 분석은 향후 3D-HEVC 기술에 대한 알고리즘 개발을 위한 기술 선별 및 조정에 유용할 것으로 판단된다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.29 no.2C
• /
• pp.315-323
• /
• 2004

In this paper we present a method to generate the bit stream of static video test patterns for testing the picture quality in DTV system. The proposed user-defined quantization table is suitable for the static video test pattern and for minimizing the deterioration of picture quality by quantization, the underflow or overflow of video buffer generated on the process of coding the static video test pattern is compensated by a adaptive zero stuffing algorithm so that optimal picture quality is implemented. Experimental result showed that the test pattern stream encoded by MPEG-2 software with the proposed algorithm had a stable bit rate and good video quality during the decoding process, which is about 3 dB higher than that of the conventional case.

• Journal of Broadcast Engineering
• /
• v.21 no.6
• /
• pp.878-888
• /
• 2016

This paper proposes an efficient method to remove the boundary artifacts of rendered views caused by damaged depth maps in the 3D video system. First, characteristics of boundary artifacts with the compression noise in depth maps are carefully studied. Then, the artifacts suppression method is proposed by the iterative projection onto convex sets (POCS) algorithm with setting the convex set in pixel and frequency domain. The proposed method is applied to both texture and depth maps separately during view rendering. The simulation results show the boundary artifacts are greatly reduced with improving the quality of synthesized views.

• Journal of Broadcast Engineering
• /
• v.17 no.6
• /
• pp.1004-1013
• /
• 2012

3D images and videos are required viewers to wear 3D glasses. According to the data, about half of Korean people wear glasses or contact lens and this implies 3D video viewers may have a trouble due to putting a pair of 3D glasses atop their glasses. The purpose of this study is to examine gamers' perceived characteristics, presence, and fatigue according to video gaming display (2D vs. 3D) and glasses whether wearing or not. The results show that the interaction effect of the display and wearing glasses was statistically significant in the perceived presence, and the main effect of the display was statistically significant in the perceived characteristics and fatigue.

• Journal of KIISE:Computer Systems and Theory
• /
• v.33 no.6
• /
• pp.354-362
• /
• 2006

This paper suggests a method for generating high qualify stereo video avatar to support visual communication in a CAVE$^{TM}$-like system. In such a system because of frequent change of light projected onto screens around user, it is not easy to extract user silhouette robustly, which is an essential step to generate a video avatar. In this study, we use an infrared reflective image acquired by a grayscale camera with a longpass filter so that the change of visible light on a screen is blocked to extract robust user silhouette. In addition, using two color cameras positioned at a distance of a binocular disparity of human eyes, we acquire two stereo images of the user for fast generation and stereoscopic display of a high quality video avatar without 3D reconstruction. We also suggest a fitting algorithm of a silhouette mask on an infrared reflective image into an acquired color image to remove background. Generated stereo images of a video avatar are texture mapped into a plane in virtual world and can be displayed in stereoscopic using frame sequential stereo method. Suggested method have advantages that it generates high quality video avatar taster than 3D approach and it gives stereoscopic feeling to a user 2D based approach can not provide.

• ETRI Journal
• /
• v.34 no.5
• /
• pp.666-673
• /
• 2012

Non-real-time delivery of stereoscopic video has been considered as a service scenario for 3DTV to overcome the limited bandwidth in the terrestrial digital television system. A hybrid codec combining MPEG-2 and H.264/AVC has been suggested for the compression of stereoscopic video for 3DTV. In this paper, we propose a stereoscopic video coding scheme using adaptive pre-/post-filters (APPF) to improve the quality of 3D video while retaining compatibility with legacy video coding standards. The APPF are applied adaptively to blocks of various sizes determined by the macroblock coding mode and reference frame index. Experiment results show that the proposed method achieves up to 24.86% bit rate savings relative to a hybrid codec of MPEG-2 and H.264/AVC including the inter-view prediction.