• Journal of Institute of Control, Robotics and Systems
• /
• v.10 no.1
• /
• pp.54-59
• /
• 2004

We suggest a 3D vision inspection algorithm which is based on the external shape feature. Because many electronic parts have the regular shape, if we have the database of pattern and can recognize the object using the database of the object s pattern, we can inspect many types of electronic parts. Our proposed algorithm uses the geometrical pattern matching method and 3D database on the electronic parts. We applied our suggested algorithm fer inspecting several objects including typical IC and capacitor. Through the experiments, we could find that our suggested algorithm is more effective and more robust to the inspection environment(rotation angle, light source, etc.) than conventional 2D inspection methods. We also compared our suggested algorithm with the feature space trajectory method.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.19 no.2
• /
• pp.335-343
• /
• 2024

Real-time target signal processing techniques for FMCW LiDAR are described in this paper. FMCW LiDAR is gaining attention as the next-generation LiDAR for self-driving cars because of its detection robustness even in adverse environmental conditions such as rain, snow and fog etc. in addition to its long range measurement capability. The hardware architecture which is required for high-speed data acquisition, data transfer, and parallel signal processing for frequency-domain signal processing is described in this article. Fourier transformation of the acquired time-domain signal is implemented on FPGA in real time. The paper also details the C-FAR algorithm for ensuring robust target detection from the transformed target spectrum. This paper elaborates on enhancing frequency measurement resolution from the target spectrum and converting them into range and velocity data. The 3D image was generated and displayed using the 2D scanner position and target distance data. Real-time target signal processing and high-resolution image acquisition capability of FMCW LiDAR by using the proposed parallel signal processing algorithms based on FPGA architecture are verified in this paper.

• Proceedings of the IEEK Conference
• /
• 2003.07e
• /
• pp.1907-1910
• /
• 2003

This paper describes a method of stereo image composition for mixed reality without camera calibration or complicate tracking algorithm. The proposed system tracks the panel which has blob makers, and composes virtual objects naturally using the method of texture mapping which is often used in geological computer graphics mapping when we do mapping 2D computer graphic data or man-made 2D images. The proposed algorithm makes it possible for us to compose virtual data even in the case that the panel is bent. For composing 3D object, the system uses depth information obtained from stereo image so that we do not need cumbersome procedure of camera calibration.

• The Journal of the Korea Contents Association
• /
• v.10 no.6
• /
• pp.222-235
• /
• 2010

After its remarkable success, "Avatar" brought another way of creating cinematic story; the stereoscopic 3-D cinema. Comparing conventional 2-D filmmaking, you need twice as much of budget and manpower in 3-D filmmaking because of the complicated process and slow production speed. The 3-D hardware like 3-D TV and 3-D projector are already showing at the retail stores while 3-D filmmaking is still in veil, and no major educational institution is yet to start 3-D related education. As 3-D movies get popular and demand more 3-D filmmaking professional crew, educating 3-D filmmaking to 2-D based film students will improve their hiring rate in the market. The successful result of the box office showing 3-D films like "Alice in Wonderland 3D" and "Titan 3D" forecasts that there will be more demand on 3-D related jobs very soon.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.8 no.12
• /
• pp.1933-1940
• /
• 2013

Recently, due to improvement of computer performance and development of information devices, technology of mobile augmented reality is proliferating rapidly. However, because most of contents are passive or limitary, user can not feel interest and immersion. This paper designs interaction user interface system of 2 dimensional drawing based on mobile augmented reality to make bi-directional communication between the real world and the virtual world possible by using the vision based augmented reality and the database system.

• Journal of Korean Society of Forest Science
• /
• v.109 no.1
• /
• pp.23-30
• /
• 2020

The use of increased computing power, machine learning, and deep learning techniques have dramatically increased in various sectors. In particular, image detection algorithms are broadly used in forestry and remote sensing areas to identify forest types and tree species. However, in South Korea, machine learning has rarely, if ever, been applied in forestry image detection, especially to classify tree species. This study integrates the application of machine learning and forest image detection; specifically, we compared the ability of two machine learning data collection methods, namely image data captured by forest experts (D1) and web-crawling (D2), to automate the classification of five trees species. In addition, two methods of characterization to train/test the system were investigated. The results indicated a significant difference in classification accuracy between D1 and D2: the classification accuracy of D1 was higher than that of D2. In order to increase the classification accuracy of D2, additional data filtering techniques were required to reduce the noise of uncensored image data.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2012.07a
• /
• pp.451-452
• /
• 2012

PDP, LCD, IED 등의 디스플레이 패널 재료를 회두로 생장해 오던 디스플레이 장치 시장이 최근 3D라는 어플리케이션이 화두가 되어 시장 성장을 이끌어 오고 있다. 3D 디스플레이 장치 시장의 규모 및 성장속도에 비해 이를 이용할 수 있도록 하는 3D 서비스는 상대적으로 부족하다. 3D 디스플레이 장치를 확보하는 소비자가 늘어남에 따라 지속적으로 증가하는 3D 서비스에 대한 잠재 수요에 대응하기 위해 다양한 형태의 3D 서비스가 계획되고, 다양한 형태의 콘텐츠 제작이 이루어지고 있다. 3D 디스플레이 장치 및 서비스 시장의 지속적인 성장을 위해서는 제작 및 서비스되는 3D 콘텐츠에 대한 품질 평가 방법에 대한 연구가 선행되어야 한다. 이에 본 논문은 3D 비디오에 대해 사람들이 느끼는 주관적 품질 평가 방법을 제안한다. 하나의 영상 도메인에 표현되던 2D 영상과 달리 스테레오스코픽 비전 방식의 3D 입체 영상은 오른쪽과 왼쪽, 두 개의 영상 도메인에 각 각 표현된다. 또한 양쪽 영상의 차이로 인해 깊이감을 인식할 수 있게 된다. 본 논문에서는 양안에 입력되는 영상의 품질을 각각 측정하고, 두 영상의 차이가 이끌어내는 깊이감을 분석하여 입체영상 전체의 품질 평가에 이용하는 방법을 제안한다.

• Journal of Broadcast Engineering
• /
• v.24 no.5
• /
• pp.765-774
• /
• 2019

In this paper, we propose a point cloud matching algorithm for multiple RGB-D cameras. In general, computer vision is concerned with the problem of precisely estimating camera position. Existing 3D model generation methods require a large number of cameras or expensive 3D cameras. In addition, the conventional method of obtaining the camera external parameters through the two-dimensional image has a large estimation error. In this paper, we propose a method to obtain coordinate transformation parameters with an error within a valid range by using depth image and function optimization method to generate omni-directional three-dimensional model using 8 low-cost RGB-D cameras.

• Proceedings of the Korean Information Science Society Conference
• /
• 2008.06c
• /
• pp.487-491
• /
• 2008

카메라로부터 얻어지는 화상정보를 처리하여 사람을 검지하는 기술은 많은 분야에 적용될 수 있다. 실제로 많은 어플리케이션에 적용되고 있다. 현재 Tracking 기술에 관련한 다양한 논문과 방법들이 존재한다. 본 논문에서는 스테레오비전이 아닌 2-D조건에서 움직이는 물체와 움직이지 않는 물체를 구분하여, 구분된 영역에서 탬플릿 매칭을 통하여 사람 검지여부를 결정하는 알고리즘을 제안한다.

• Annual Conference of KIPS
• /
• 2016.04a
• /
• pp.706-707
• /
• 2016

본 논문은 HMD 환경에서 스마트 디바이스를 입력장치로 이용하여 증강 객체의 동적 / 정적 속성을 부여하는 증강현실 콘텐트 저작 시스템을 제안한다. 본 논문에서는 2D 터치 스크린을 이용하여 증강 객체의 3차원 위치, 자세, 크기를 변경하는 방법을 제안한다. 또한 증강 객체를 이용해 객체의 동적 경로를 설정하여 객체의 움직임을 표현한다. 제안하는 시스템을 통해 비전문가도 현장에서 즉석으로 동적 / 정적 속성을 갖는 증강 객체 콘텐트 저작이 가능하다.