-
公开(公告)号：KR101586071B1
公开(公告)日：2016-02-03
申请号：KR1020120095008
申请日：2012-08-29
Applicant: 한국과학기술연구원
Abstract: 본발명은, 크기가큰 대상객체의전체형상을촬영한참조영상을분할하여분할영상별로특징점을학습하고, 이를실시간영상의특징점과비교하여추정한실시간영상의촬영자세및 위치를기초로증강현실서비스를제공하는마커리스증강현실서비스제공장치및 이를위한촬영위치추정방법에관한것으로, 본발명에따르면, 종래에는마커리스증강현실서비스를제공하기어려울정도로크기가큰 대형물체를대상객체로하여실시간으로촬영자세및 위치를추정하여증강현실서비스를제공할수 있게된다. 이는스마트폰이나태블릿 PC와같은소형의단말장치에서증강현실서비스를제공함에있어서특히유용하다.
-
公开(公告)号：KR1020140030444A
公开(公告)日：2014-03-12
申请号：KR1020120095008
申请日：2012-08-29
Applicant: 한국과학기술연구원
Abstract: The present invention relates to a marker-less augmented reality service providing apparatus and a photographing position estimating method therefor, which studies the feature point of each divided image by dividing a reference image which photographs the entire shape of a target object of a big size and provides an augmented reality service based on the photographing posture and position of a real-time image which is estimated by being compared with the feature point of the real-time image. According to the present invention, the augmented reality service can be provided by estimating the posture and position of a photograph in real-time by having an object which is large provide the mark-less augmented reality service in an existing service as a target object. The present invention is especially useful when providing an augmented reality service in a small terminal device such as a smartphone or a tablet PC. [Reference numerals] (AA) Start; (BB) End; (S301) Divide a reference image; (S302) Extract the feature point of each of the divided images; (S303) Store the extracted feature point according to the divided images; (S304) Obtain a real-time image; (S305) Extract the feature point of the real-time image; (S306) Match the feature point of the divided images with the feature point of the real-time image; (S307) Extract a matched divided image which is matched with the real-time image based on the matched result; (S308) Estimate the photographing posture and position of the real-time image based on the relative position of a feature point which is matched with the feature point of the real-time image in the divided image which is extracted by the matching
Abstract translation: 无标记增强现实服务提供装置及其拍摄位置估计方法技术领域本发明涉及一种无标记增强现实服务提供装置及其拍摄位置估计方法，该方法通过划分拍摄大尺寸目标对象的整体形状的参考图像，以及 基于通过与实时图像的特征点进行比较而估计的实时图像的拍摄姿态和位置，提供增强现实服务。 根据本发明，可以通过使现有服务中的无标记增强现实服务作为目标对象来实时估计照片的姿势和位置来提供增强现实服务，以提供无标记增强现实服务。 当在诸如智能电话或平板电脑的小型终端设备中提供增强现实服务时，本发明特别有用。 （附图标记）（AA）开始; （BB）结束; （S301）分割参照图像; （S302）提取每个分割图像的特征点; （S303）根据划分的图像存储提取出的特征点; （S304）获取实时图像; （S305）提取实时图像的特征点; （S306）将分割图像的特征点与实时图像的特征点进行匹配; （S307）基于匹配结果提取与实时图像匹配的匹配分割图像; （S308）基于与通过匹配的提取的分割图像中的实时图像的特征点匹配的特征点的相对位置来估计实时图像的拍摄姿势和位置
-
公开(公告)号：KR101439406B1
公开(公告)日：2014-09-15
申请号：KR1020120085116
申请日：2012-08-03
Applicant: 한국과학기술연구원
IPC: G06T17/20
Abstract: 본 발명의 일 실시예에 따르는 영상처리방법은 영상의 복수의 픽셀에 대응하는 복수의 정점(vertexes)과 복수의 다각형으로 구성되는 평면 메쉬를 형성하는 단계; 상기 각 정점과 대응하는 각 픽셀의 밝기 값에 상응하는 높이를 상기 복수의 정점에 설정하여 입체 메쉬를 형성하는 단계; 상기 높이의 차이가 기 설정된 범위 내에 있는 메쉬 영역을 결합하여 상기 입체 메쉬를 단순화하는 단계; 상기 영상의 깊이 값들을 갖는 복수의 기준점을 상기 영상의 시점(view point)으로 상기 영상에 매칭하는 단계; 상기 매칭된 기준점에 상기 단순화된 입체 메쉬의 정점들을 대응시켜 상기 정점들의 깊이 값을 추출하는 단계;를 포함하는 것을 특징으로 한다.
-
公开(公告)号：KR1020140010844A
公开(公告)日：2014-01-27
申请号：KR1020120085116
申请日：2012-08-03
Applicant: 한국과학기술연구원
IPC: G06T17/20
Abstract: An image processing method according to an embodiment of the present invention comprises the steps of: forming a planar mesh composed of a plurality of vertexes corresponding to a plurality of pixels of an image and a plurality of polygons; setting, to the vertexes, a height corresponding to a brightness value of each pixel corresponding to each vertex to form a three-dimensional mesh; combining mesh areas where the difference among the heights is within a preset range to simplify the three-dimensional mesh; matching, to the image, a plurality of reference points having depth values of the image as the viewpoint of the image; and matching the vertexes of the simplified three-dimensional mesh to the matched reference points to extract the depth values of the vertexes. [Reference numerals] (AA) Start; (BB) End; (S10) Planar mesh generation; (S20) Three-dimensional mesh generation; (S30) Three-dimensional mesh simplification; (S40) Reference point matching; (S50) Depth information extraction; (S60) Depth information adjustment
Abstract translation: 根据本发明的实施例的图像处理方法包括以下步骤：形成由与图像的多个像素对应的多个顶点和多个多边形组成的平面网格; 对顶点设置与每个顶点对应的每个像素的亮度值对应的高度，以形成三维网格; 组合高度差在预设范围内的网格区域，以简化三维网格; 与图像匹配多个参考点，其具有作为图像的视点的图像的深度值; 并将简化的三维网格的顶点与匹配的参考点相匹配，以提取顶点的深度值。 （附图标记）（AA）开始; （BB）结束; （S10）平面网格生成; （S20）三维网格生成; （S30）三维网格简化; （S40）参考点匹配; （S50）深度信息提取; （S60）深度信息调整
-
-
-
Search Results
4
records
(took 0.018 seconds)
