-
公开(公告)号：KR1020090064243A
公开(公告)日：2009-06-18
申请号：KR1020070131827
申请日：2007-12-15
Applicant: 한국전자통신연구원
IPC: H04N9/64
CPC classification number: H04N17/02 , G06T7/80 , H04N13/133 , H04N13/239 , H04N13/243 , H04N13/246 , H04N13/25 , H04N13/257 , H04N17/002
Abstract: A color equalizing method between multiple cameras using a color checker chart is provided to be useful to 3D information restoration of a dynamic object, in a circumstance where it is not possible to perform a photographing process as moving one camera. In case of 3D data restoration, color differences generated between images simultaneously taken in various directions from plural cameras are calculated(S200-S202). With regard to the color differences, a relation between color models of each camera and a reference color model of a color checker cart is modeled and corrected to equalize colors between the multiple cameras(S203-S205). Coefficients of a color mapping equation between input image color information of the cameras and reference color information of the color checker chart are estimated. The estimated coefficients are applied to equalize input image color values of the cameras.
Abstract translation: 在使用彩色检查器图表的多个摄像机之间的色彩均衡方法被提供用于在不可能作为移动一个摄像机执行拍摄处理的情况下用于动态对象的3D信息恢复。 在3D数据恢复的情况下，计算从多个摄像机在各个方向上同时拍摄的图像之间产生的色差（S200-S202）。 关于颜色差异，对每个照相机的颜色模型和颜色检查器车的参考颜色模型之间的关系进行建模和校正以均衡多个照相机之间的颜色（S203-S205）。 估计摄像机的输入图像颜色信息和颜色检查图表的参考颜色信息之间的颜色映射方程的系数。 应用估计的系数来均衡摄像机的输入图像颜色值。
-
公开(公告)号：KR1020090064153A
公开(公告)日：2009-06-18
申请号：KR1020070131729
申请日：2007-12-15
Applicant: 한국전자통신연구원
Abstract: A 3D facial model generation method and a device using multi-viewpoint image information are provided to control expression or facial postures through join control of a hierarchical structure, thereby offering an easy editing and modifying function of external facial shapes or expressions. An image input unit(102) inputs multi-viewpoint image information toward an expressionless face of a performer captured from plural cameras. An outer shape restorer(104) generates a mesh structure for outer shape expressions of a 3D facial model by using the inputted multi-viewpoint image information. A standard facial model transition unit(106) transfers standard model information to the generated mesh structure, and planarizes the mesh structure. The standard facial model transition unit enables operation of moving parts in each part. An intrinsic expression learning unit(108) locates joints of a hierarchical structure in the mesh structure.
Abstract translation: 提供3D面部模型生成方法和使用多视点图像信息的装置，以通过层次结构的连接控制来控制表情或面部姿势，从而提供外部面部形状或表情的容易的编辑和修改功能。 图像输入单元（102）将多视点图像信息输入到从多个照相机拍摄的表演者的无表情面。 外部形状恢复器（104）通过使用输入的多视点图像信息生成3D面部模型的外部形状表达的网格结构。 标准面部模型转换单元（106）将标准模型信息传送到生成的网格结构，并且平面化网格结构。 标准面部模型转换单元可以在每个部件中运动部件。 内在表达式学习单元（108）将网格结构中的层次结构的关节定位。
-
公开(公告)号：KR100871149B1
公开(公告)日：2008-12-05
申请号：KR1020070100389
申请日：2007-10-05
Applicant: 한국전자통신연구원
IPC: H04N5/232
Abstract: 본 발명은 영상 간의 대응점 위치로부터 고정된 줌 렌즈 카메라 초점거리를 계산하는 카메라 자동보정 방법에 관한 것으로, 초점거리를 변수로 하는 비용함수(cost function)를 정의하고, 정의한 비용함수를 최소화하는 초점거리를 구하여 모든 영상 쌍 각각에서 계산되는 대응점의 3차원 복원결과가 최대한 일치하는 초점거리를 구함으로써, 계산된 초점거리의 신뢰도를 쉽게 검증할 수 있을 뿐 아니라, 입력 대응점의 위치가 부정확하게 주어진 경우에도 안정적으로 카메라 초점거리를 계산할 수 있다.
카메라 보정(camera calibration), 카메라 자동보정(camera self-calibration), 카메라 내부인자(camera intrinsic parameter), 카메라 초점거리 (camera focal length), 3차원 복원(3D reconstruction), 카메라 자세추정(camera motin estimation)-
公开(公告)号：KR1020080051007A
公开(公告)日：2008-06-10
申请号：KR1020070094263
申请日：2007-09-17
Applicant: 한국전자통신연구원
CPC classification number: G06T17/205 , G06T13/40 , G06T15/04 , G06T17/10
Abstract: A system and a method for generating a 3D face modeling and animation using a single video camera are provided to transfer facial animation data of an actor into a facial performance of a virtual actor. A video camera(10) captures an image of an actor to generate video. A computer(20) processes the video of the actor inputted via the video camera and animate it. A digital signal processor(22) receives a facial performance video of a user from the video camera, converts the 3D facial performance video in an analog format into digital signals expressed by 0 and 1, so that the digital signals can be processed according to a mathematical operation. A face modeling unit(24) receives the video of the actor and models 3D face of the actor by frames. A D/A converter(28) converts the animation image of the 3D face processed by the face modeling unit into analog signals so as to be displayed via an output unit(30).
Abstract translation: 提供了使用单个摄像机生成3D脸部建模和动画的系统和方法，以将演员的面部动画数据转换成虚拟演员的面部表演。 视频摄像机（10）捕获演员的图像以产生视频。 计算机（20）处理通过摄像机输入的动作的视频并对其进行动画处理。 数字信号处理器（22）从视频摄像机接收用户的面部表情视频，将模拟格式的3D面部表情视频转换成由0和1表示的数字信号，使得数字信号可以根据 数学运算。 面部建模单元（24）接收演员的视频并且通过帧模拟演员的3D面部。 D / A转换器（28）将由脸部建模单元处理的3D脸部的动画图像转换为模拟信号，以经由输出单元（30）显示。
-
公开(公告)号：KR1020050058659A
公开(公告)日：2005-06-17
申请号：KR1020030090603
申请日：2003-12-12
Applicant: 한국전자통신연구원
IPC: H04N13/00
Abstract: 본 발명은 방향 필터링을 이용한 스테레오 영상의 정합 시스템 및 그 방법에 관한 것으로, 특히 본 발명의 방법은 두 개의 영상으로부터 화소 사이의 밝기값 차이의 크기(DS)를 계산하는 단계와, DS값을 서로 다른 방향을 가지는 n(n>2)개의 방향 필터를 사용하여 필터링하는 단계와, DS값을 합산 필터링하여 합산 필터링 연산값(S(i, j, z))을 구하고 최소화 필터링으로 합산 필터링 연산값(S(i, j, z))들 중에서 최소값MF(i, j, z)을 구하는 단계와, 방향 필터링된 값(
-
Patent Agency Ranking
公开(公告)号：KR1020040049526A
公开(公告)日：2004-06-12
申请号：KR1020020077324
申请日：2002-12-06
Applicant: 한국전자통신연구원
CPC classification number: G06T13/40
Abstract: PURPOSE: A system for controlling a virtual character is provided to configure a virtual character with one mesh, and to generate motions of joint parts as a weighted average of motions of peripheral joints in a mesh skinning method, thereby realizing a very natural motion of the virtual character. CONSTITUTION: A multi-mesh structure storage(100) stores a character shape mesh interworked with joint information of a character and at least one item mesh interworked with the joint information in assembled type. A character motion controller(300) performs a mesh skinning process according to character motion control information, and controls the item mesh and the shape mesh attached to the character among the at least one item mesh. A temporary mesh generator(200) synthesizes the item mesh attached to the character with the shape mesh, and generates a temporary mesh.
Abstract translation: 目的：提供一种用于控制虚拟角色的系统，以一个网格配置一个虚拟角色，并且以网格去皮方式生成关节部分的运动作为外围关节运动的加权平均值，从而实现一个非常自然的运动 虚拟角色 构成：多网格结构存储（100）存储与字符的联合信息相互配合的字符形状网格以及与组合类型的联合信息相互配合的至少一个项目网格。 角色运动控制器（300）根据角色运动控制信息执行网格去皮处理，并且控制在至少一个项目网格中附加到角色的项目网格和形状网格。 临时网格生成器（200）将与字符连接的项目网格与形状网格合成，并生成临时网格。
-
-
公开(公告)号：KR101841668B1
公开(公告)日：2018-03-27
申请号：KR1020120015325
申请日：2012-02-15
Applicant: 한국전자통신연구원
IPC: G06T17/00
CPC classification number: G06T17/00 , G06T2210/56
Abstract: 본발명은 3차원스캐닝을통해획득한포인트클라우드와영상을이용하여정적인배경의 3차원모델을생성하는입체모델생성장치및 방법을제안한다. 이를위해본 발명은미리정해진지역을촬영하여얻은카메라영상에상기지역을스캐닝하여얻은포인트클라우드(point cloud)를정합하여정합영상을생성하는영상정합부; 상기지역에위치하는오브젝트를메쉬모델로생성하는메쉬모델생성부; 및상기메쉬모델에상기정합영상으로부터얻은텍스처정보를반영하여상기오브젝트에대한입체모델을생성하는입체모델생성부를포함하는것을특징으로하는입체모델생성장치를제안한다. 본발명은 3차원지도서비스에이용될수 있다.
Abstract translation: 本发明提出了一种用于使用点云和通过三维扫描获得的图像来生成静态背景的三维模型的三维模型生成设备和方法。 为此，本发明提供了一种图像处理装置，包括：图像匹配单元，用于匹配通过扫描通过拍摄预定区域而获得的照相机图像上的区域而获得的点云，以生成匹配图像; 网格模型生成单元，其生成位于该区域中的对象作为网格模型; 以及三维模型生成单元，用于通过将从匹配图像获得的纹理信息反映到网格模型来生成对象的三维模型。 本发明可以用于三维地图服务。
-
公开(公告)号：KR1020160058563A
公开(公告)日：2016-05-25
申请号：KR1020140160222
申请日：2014-11-17
Applicant: 한국전자통신연구원
IPC: G06T17/20
CPC classification number: G06T17/20 , G06T17/205 , Y02T10/82
Abstract: 본발명은 3차원프린팅시에발생할수 있는 3차원물체의파손을최소화시키기위해 3차원물체의내구성을사전에검증하는방법및 장치에관한것이다. 본발명에따른 3차원메쉬검증방법은 3차원메쉬데이터파일을분석하여상기 3차원메쉬의취약부분을산출하는단계; 및상기산출된 3차원메쉬의취약부분을모니터에표시하고, 사용자로하여금상기취약부분을수정할수 있도록해주는단계를포함한다.
Abstract translation: 本发明涉及一种用于验证3D对象的耐久性的方法和装置，以便最小化在3D打印期间可能发生的3D对象的损坏。 根据本发明的实施例，用于验证3D网格的方法包括以下步骤：分析3D网格数据文件以产生3D网格的弱点; 并且在监视器上显示所产生的3D网格的弱点，并允许用户校正弱点。
-
公开(公告)号：KR101575284B1
公开(公告)日：2015-12-09
申请号：KR1020110023508
申请日：2011-03-16
Applicant: 한국전자통신연구원
IPC: G06T15/04
Abstract: 본발명은 2차원영상의해상도레벨에따른 3차원복원물체의텍스처생성장치및 그방법에관한것으로, 거리가다른적어도둘 이상의영역에서촬영된영상들로부터 3차원물체정보및 영상들을촬영한카메라정보를추출하여영상들내에포함된 3차원물체를복원하는 3차원물체복원부, 영상들의한 픽셀이 3차원물체의실사영상에서차지하는공간영역의크기를측정하여영상들에대한해상도를산출하는해상도산출부, 해상도의레벨별로분류된각각의영상들을이용하여영상들에대응하는레벨별텍스처를생성하는텍스처생성부, 및 3차원물체의화면상의크기에따라해당되는레벨의텍스처를선택하여렌더링을수행하는렌더링부를포함한다.
-
-
-
-
-
-
-
-
-
Search Results
94
records
(took 0.024 seconds)
