• International journal of advanced smart convergence
• /
• v.9 no.3
• /
• pp.192-198
• /
• 2020

The 3D point cloud is a key technology of object detection for virtual reality and augmented reality. In order to apply various areas of object detection, it is necessary to obtain 3D information and even color information more easily. In general, to generate a 3D point cloud, it is acquired using an expensive scanner device. However, 3D and characteristic information such as RGB and depth can be easily obtained in a mobile device. GNN (Graph Neural Network) can be used for object detection based on these characteristics. In this paper, we have generated RGB and RGBD by detecting basic information and characteristic information from the KITTI dataset, which is often used in 3D point cloud object detection. We have generated RGB-GNN with i-GNN, which is the most widely used LiDAR characteristic information, and color information characteristics that can be obtained from mobile devices. We compared and analyzed object detection accuracy using RGBD-GNN, which characterizes color and depth information.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.06a
• /
• pp.1241-1244
• /
• 2022

최근 K-Pop 을 위시한 예술공연 콘텐츠에 몰입형 미디어를 접목한 온택트 (Ontact) 미디어 스트리밍 서비스가 주목받고 있는 가운데, 본 논문은 일반적으로 사용되는 2D 디스플레이 또는 HMD (Head-Mounted Display) 기반 VR (Virtual Reality, VR) 서비스에서 탈피하여, 대형 가상현실 공연장을 위한 360 도 VR 비디오 스트리밍 시스템을 제안한다. 제안된 시스템은 Phase 1, 2, 3 의 연구개발 단계를 밟아 6DoF (Degrees of Freedom) 시점 자유도를 지원하는 360 도 VR 비디오 스트리밍 시스템을 개발하는 것을 최종목표로 하고 있으며, 현재는 Phase 1: 대형 가상현실 공연장을 위한 3DoF 360 도 VR 비디오 스트리밍 시스템 프로토타입의 개발까지 완료되었다. 구현된 스트리밍 시스템 프로토타입은 서브픽처 기반 Viewport-dependent 스트리밍 기술이 적용되어 있으며, 기존 방식과 비교하였을 때 약 80%의 비트율 감소, 약 543%의 영상 디코딩 속도 향상을 확인하였다. 또한, 단순 구현 및 성능평가에서 그치지 않고, 실제 미국 UCSB 에 위치한 대형 가상현실 공연장 AlloSphere 에서의 시범방송을 수행하여, 향후 Phase 2, 3 연구단계를 위한 연구적 기반을 마련하였다.

• Journal of Broadcast Engineering
• /
• v.11 no.4 s.33
• /
• pp.441-447
• /
• 2006

See-through HMD type 3D display can provide an advantage of us seeing virtual 3D data used stereoscopic display simultaneously with real object(MR-Mixed Reality). But, when user sees stereoscopic display for a long time, not only eye fatigue phenomenon happens but also de-focus phenomenon of data happens by fixed focal point of virtual data. Dissatisfaction of focus adjustment of eye can be considered as the important reason of this phenomenon. In this paper, We proposed an application of multi-focus in see-through HMD as a solution of this problem. As a result, we confirmed that the focus adjustment coincide between the object of real world and the virtual data by multi-focus in monocular condition.

• Journal of Broadcast Engineering
• /
• v.15 no.5
• /
• pp.607-616
• /
• 2010

The methods developed to measure visual fatigue so far are quite few and lack of validity, and more importantly, they do not figure out the complex properties of the visual fatigue. The purpose of the research was to analyze the factors comprising the visual fatigue and to develop the method to measure it validly. The results are summarized as follows. First, we found that the 3D visual fatigue was comprised of four independent factors (visual stress, eye pain, body pain and image blurring factors). Second, we developed 29 items that measure four factors of 3D visual fatigue. Finally, the watching duration and binocular disparities affected the visual fatigue as had been expected. These results imply that the developed method does measure the three dimensional fatigue validly.

• Journal of Broadcast Engineering
• /
• v.17 no.4
• /
• pp.625-633
• /
• 2012

Extended Depth of Focus (EDF) algorithms for extracting three-dimensional (3D) information from a set of optical image slices are studied by many researches recently. Due to the limited depth of focus of the microscope, only a small portion of the image slices are in focus. Most of the EDF algorithms try to find the in-focus area to generate a single focused image and a 3D depth image. Inherent to most image processing algorithms, the EDF algorithms need parameters to be properly initialized to perform successfully. In this paper, we select three popular transform-based EDF algorithms which are each based on pyramid, wavelet transform, and complex wavelet transform, and study the performance of the algorithms according to the initialization of its parameters. The parameters we considered consist of the number of levels used in the transform, the selection of the lowest level image, the window size used in high frequency filter, the noise reduction method, etc. Through extended simulation, we find a good relationship between the initialization of the parameters and the properties of both the texture and 3D ground truth images. Typically, we find that a proper initialization of the parameters improve the algorithm performance 3dB ~ 19dB over a default initialization in recovering the 3D information.

• International journal of advanced smart convergence
• /
• v.8 no.3
• /
• pp.39-45
• /
• 2019

This paper presents an optical technique for the three-dimensional (3D) shape inspection of micro solder balls used in ball-grid array (BGA) packaging. The proposed technique uses an optical source composed of spatially arranged light-emitting diodes (LEDs) and the results are derived based on the specular reflection characteristics of the micro solder balls for BGA A vision system comprising a camera and LEDs is designed to capture the reflected images of multiple solder balls arranged arbitrarily on a tray and the locations of the LED point-light-source reflections in each ball are determined via image processing, for shape inspection. The proposed methodology aims to determine the presence of defects in 3D BGA shape using the statistical information of the relative positions of multiple BGA balls, which are included in the image. The presence of the BGA balls with large deviations in relative position imply the inconsistencies in their shape. Experiments were conducted to verify that the proposed method could be applied to inspection without sophisticated mechanism and productivity problem.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.21 no.3
• /
• pp.59-66
• /
• 2021

Humans mainly recognize surrounding objects using visual and auditory information among the five senses (sight, hearing, smell, touch, taste). Major research related to the latest object recognition mainly focuses on analysis using image sensor information. In this paper, after emitting various chirp audio signals into the observation space, collecting echoes through a 2-channel receiving sensor, converting them into spectral images, an object recognition experiment in 3D space was conducted using an image learning algorithm based on deep learning. Through this experiment, the experiment was conducted in a situation where there is noise and echo generated in a general indoor environment, not in the ideal condition of an anechoic room, and the object recognition through echo was able to estimate the position of the object with 83% accuracy. In addition, it was possible to obtain visual information through sound through learning of 3D sound by mapping the inference result to the observation space and the 3D sound spatial signal and outputting it as sound. This means that the use of various echo information along with image information is required for object recognition research, and it is thought that this technology can be used for augmented reality through 3D sound.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.21 no.4
• /
• pp.443-451
• /
• 2010

This paper describes a fully integrated CMOS low-IF mobile-TV RF tuner for Band-III T-DMB/DAB applications. All functional blocks such as low noise amplifier, mixers, variable gain amplifiers, channel filter, phase locked loop, voltage controlled oscillator and PLL loop filter are integrated. The gain of LNA can be controlled from -10 dB to +15 dB with 4-step resolutions. This provides a high signal-to-noise ratio and high linearity performance at a certain power level of RF input because LNA has a small gain variance. For further improving the linearity and noise performance we have proposed the RF VGA exploiting Schmoock's technique and the mixer with current bleeding, which injects directly the charges to the transconductance stage. The chip is fabricated in a 0.18 um mixed signal CMOS process. The measured gain range of the receiver is -25~+88 dB, the overall noise figure(NF) is 4.02~5.13 dB over the whole T-DMB band of 174~240 MHz, and the measured IIP3 is +2.3 dBm at low gain mode. The tuner rejects the image signal over maximum 63.4 dB. The power consumption is 54 mW at 1.8 V supply voltage. The chip area is $3.0{\times}2.5mm^2$.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2017.06a
• /
• pp.248-249
• /
• 2017

최근 들어, 실감형 콘텐츠를 제작하기 위한 무안경 다시점 3DTV 와 가상, 증강현실 콘텐츠에 대한 기술개발이 활발해 지고 있다. 또한, UHDTV 의 등장으로 무안경 다시점 3DTV 의 걸림돌이던 해상도 저하문제가 해결되면서 시점당 해상도가 향상된 다시점 3D 디스플레이 상용품 및 시제품들이 발표되고 있다. 하지만, 시청 위치가 한정되어 있고, 시청자에게 눈의 피로를 일으키며, 비싼 가격으로 인해 아직까지 소비자에게 좋은 반응을 얻지 못하고 있다. 이러한 단점을 해결하기 위해, 본 논문은 다시점 영상을 증강현실로 보여주는 방법을 소개한다. 제안한 방법을 사용하기 위해, 다시점 카메라 시스템으로 영상을 획득하고, HMD (head mounted display)와 휴대용 (handheld) 장치를 사용하여 다시점 영상을 시청한다. 실험 결과를 통해 제안한 방법이 기존의 시점 제한 문제를 해결할 수 있음을 확인했으며, 제안한 방법을 활용하여 다양한 실감형 콘텐츠를 제작할 것으로 기대한다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2011.07a
• /
• pp.22-24
• /
• 2011

본 논문에서는 3차원 디스플레이 시스템에서 다수의 중간 시점 영상을 실시간으로 생성할 수 있도록 GPU 기반의 고속 영상 합성기법을 제안하였으며 그에 대한 성능을 알아본다. 카메라의 기하 정보 및 참조 영상들의 깊이 정보를 이용하여 중간 시점 영상을 생성하였으며, 영상 합성 방법을 GPU에서 병렬 처리함으로써 고속화할 수 있었다. GPU를 효율적으로 다루기 위해 NVIDIA사의 CUDA(Compute Unified Device Architecture)$^TM$를 이용하였다. 제안한 기법은 CUDA의 SIMD(Single Instruction MUltiple Data) 구조를 사용하여 중간 영상 합성을 처리할 수 있도록 설계하였다. 본 논문은 고속 영상 합성에 중점을 두었고, 제안한 고속화 기법의 결과를 분석함으로써 다시점 3차원 디스플레이 시스템의 적용 가능성을 알아본다.