-
51.发明公开계층적인 구조의 영상 모델을 이용한 3차원 영상 생성방법과 장치, 이를 이용한 영상 인식 방법과 특징점 추출방법 그리고 상기 방법들을 수행하는 프로그램이 기록된기록매체 有权使用分层3D图像模型的3D图像生成方法和装置，使用该图像的图像识别和特征提取方法以及用于执行其方法的记录媒体存储程序
公开(公告)号：KR1020090072263A
公开(公告)日：2009-07-02
申请号：KR1020070140324
申请日：2007-12-28
Applicant: 주식회사 케이티 , 포항공과대학교 산학협력단
CPC classification number: G06T17/20 , G06T3/0093 , G06T5/50 , G06T15/04
Abstract: A 3D image generation method and an apparatus using hierarchical 3D image model, an image recognition and feature points extraction method using the same, and a recording medium storing program for performing the method thereof are provided to generate a fine 3D image, and capable of reducing processing time and operation amount for high resolution level fitting. A 2 dimensional input image is received, and a facial area is detected(110). A 3 dimensional image model of the low resolution is projected onto a 2 dimensional space(120). A first model parameter is determined(130). A 3 dimensional image model of the high resolution is projected onto the 2 dimensional space(140). A second model parameter is determined(150). A 3 dimensional image is generated by using the second model parameter(160).
Abstract translation: 提供3D图像生成方法和使用分层3D图像模型的装置，使用其的图像识别和特征点提取方法以及用于执行其方法的记录介质存储程序以产生精细的3D图像，并且能够减少 高分辨率水平拟合的处理时间和操作量。 接收二维输入图像，并检测面部区域（110）。 将低分辨率的三维图像模型投影到二维空间（120）上。 确定第一模型参数（130）。 将高分辨率的3维图像模型投影到二维空间（140）上。 确定第二模型参数（150）。 通过使用第二模型参数（160）生成3维图像。
-
公开(公告)号：KR100903096B1
公开(公告)日：2009-06-16
申请号：KR1020070063837
申请日：2007-06-27
Applicant: 포항공과대학교 산학협력단
Abstract: 본 발명에 의한 눈 검증 및 눈 위치 보정을 통한 눈 검출 방법은, 입력된 얼굴 영상에서 눈이 존재할 가능성이 있는 눈 후보 영역을 추출하는 단계; 추출한 각각의 눈 후보 영역에서 눈 검출 특징점을 통하여 눈 후보를 검출하는 단계; 각각의 눈 후보 영역에서 검출된 각각의 눈 후보와 반대편 눈 후보 영역간의 MCT 기반 상관맵에서 상관값이 피크값을 가지면 눈 후보가 정상적으로 검출된 것으로 판단하는 눈 검증 단계; 및 한쪽 눈의 후보가 잘못 검출된 경우, 정상적으로 검출된 눈과 반대편 눈 후보 영역간의 MCT 기반 상관맵에서 상관값이 피크값을 갖는 해당 좌표값으로 잘못 검출된 눈의 위치를 보정하는 단계를 포함함을 특징으로 한다.
본 발명의 눈 검출 방법은 한 쪽 눈만 정확하게 검출된 경우에도 잘못 검출된 다른 쪽의 눈의 위치를 패턴 상관맵에서 피크값의 좌표값으로 보정하여 정확하게 찾을 수 있기 때문에, 조명, 환경 변화에 매우 강인하며, 빠른 속도로 검출할 수 있는 효과를 제공한다.-
公开(公告)号：KR100779171B1
公开(公告)日：2007-11-26
申请号：KR1020060078106
申请日：2006-08-18
Applicant: 학교법인 포항공과대학교 , 포항공과대학교 산학협력단
CPC classification number: G06K9/00228 , G06T3/60 , G06T7/44 , G06T2207/30201
Abstract: An apparatus and a method for detecting a correct face in real time are provided to enable high-speed face detection by enabling a face candidate region detecting unit to extract a face candidate region by an image difference between the previous pyramid region and current pyramid region of a pyramid region in a still image or a motion image. A method for detecting a correct face in real time comprises the following steps of: extracting a face candidate region in a still image or an input image where the motion is generated; minimizing face detecting performance; detecting front, right and left faces; and detecting the correct face by minimizing the detection of a wrong face while maintaining the face detection performance by using the weight information about a face obtained in a normal face detection algorithm.
Abstract translation: 提供了一种用于实时检测正确脸部的装置和方法，用于通过使面部候选区域检测单元能够通过前一个金字塔区域和当前金字塔区域之间的图像差异提取面部候选区域来提供高速面部检测 在静止图像或运动图像中的金字塔区域。 一种用于实时检测正确脸部的方法包括以下步骤：提取静止图像中的面部候选区域或产生运动的输入图像; 最小化面部检测性能; 检测前，右，左脸; 并且通过使用关于在正常面部检测算法中获得的面部的权重信息来维持面部检测性能，通过最小化错误脸部的检测来检测正确的面部。
-
公开(公告)号：KR1020180111251A
公开(公告)日：2018-10-11
申请号：KR1020170041928
申请日：2017-03-31
Applicant: 포항공과대학교 산학협력단
IPC: G10L21/0272 , G10L21/0208 , G10L21/0356
Abstract: 소리복원방법이개시된다. 본발명의소리복원방법은카메라를통해촬영된영상으로부터분할된복수의패치를생성하는단계, 각각의패치내에포함된오브젝트의진동과관련한모션정보를이용하여, 각각의패치에대하여영상의소리를복원하는단계및, 각각의패치에대하여복원된소리들중 잡음이가장적은소리를영상의소리로선택하는단계를포함한다.
-
公开(公告)号：KR101735565B1
公开(公告)日：2017-05-15
申请号：KR1020120068176
申请日：2012-06-25
Applicant: 한화테크윈 주식회사 , 포항공과대학교 산학협력단
CPC classification number: G06T7/20 , G06T5/005 , G06T2207/10048 , G06T2207/20081 , G06T2207/30196 , G06T2207/30232
Abstract: 본발명은열 그림자를제거할수 있는움직임검출시스템및 방법에관한것으로, 본발명의일 실시예에따르면, 열그림자를제거할수 있는움직임검출시스템에있어서, 열그림자특성에대한데이터를온도별로분류및 저장하는데이터베이스; 상기열 그림자특성을기초로온도별로열 그림자의크기및 밝기정보를획득하는학습부; 검출대상열 영상에서열 그림자영역을제거하는검출부; 를포함하는움직임검출시스템이제공된다.
Abstract translation: 根据本发明的实施例，提供了一种能够去除热阴影的运动检测系统，所述系统包括： 要存储的数据库; 学习单元，用于基于热阴影特性获取每个温度的热阴影的大小和亮度信息; 检测单元，用于从检测对象热图像中去除序列阴影区域; 提供运动检测系统。
-
Patent Agency Ranking
公开(公告)号：KR101511450B1
公开(公告)日：2015-04-10
申请号：KR1020130152128
申请日：2013-12-09
Applicant: 포항공과대학교 산학협력단
CPC classification number: G06K9/6204 , G06K9/00342
Abstract: 객체의움직임을추적하는방법에있어서, 산등성이데이터추출방법과객체의관절움직임추적방법및 장치가개시된다. 산등성이데이터추출방법은, 객체의영상으로부터객체의외곽선내 포인트를선정하는단계와, 외곽선내 포인트의기울기거리값을산출하는단계와, 기울기거리값과외곽선내 포인트에인접한외곽선내 포인트들의기울기거리값의관계에기초하여외곽선내 포인트가산등성이포인트인지를결정하는단계를포함하여구성될수 있다. 객체의관절움직임추적방법은객체의산등성이포인트의분포를나타내는적어도하나의산등성이영역에관한정보를포함하는객체의초기정보를획득하는단계와, 초기정보에기반하여, 객체의시간적으로연속된영상으로부터객체의관절움직임을추적하는단계를포함하여구성될수 있다.
Abstract translation: 公开了用于提取脊数据的方法，用于跟踪物体的关节的运动的方法和装置。 用于提取脊数据的方法可以包括以下步骤：从对象的图像中选择对象的轮廓中的点; 计算轮廓中的点的倾斜距离值; 并且基于与轮廓中的点相邻的另一轮廓中的点的倾斜距离值与倾斜距离值之间的关系，确定轮廓中的点是否是脊点。 用于跟踪对象的关节运动的方法可以包括以下步骤：获得对象的初始信息，包括关于提供对象的脊点分布的至少一个脊区域的信息; 以及基于初始信息在对象的时间连续图像中跟踪对象的关节的移动。
-
公开(公告)号：KR1020140110173A
公开(公告)日：2014-09-17
申请号：KR1020130023129
申请日：2013-03-05
Applicant: 포항공과대학교 산학협력단
IPC: A61B5/103
Abstract: Disclosed are a system and a method for warning a forward head posture for preventing a herniated cervical disk. The system for warning the forward head posture includes: a posture determining device which measures the posture of a user by using an obtained image, determines a warning level based on the posture of the user and a preset reference posture, and provides the determined warning level; and a warning generating device which generates a warning corresponding to the warning level and supplies the warning level to the user. Therefore, according to the system and the method for warning the forward head posture, the user conveniently recognizes the forward head posture by additionally using a warning function in a portable external device in an environment using an information processing device. Forward head posture symptoms and the herniated cervical disk are effectively prevented by giving the warning about the posture to the user regardless of the location and kind of the information processing device.
Abstract translation: 公开了一种用于警告前头姿势以防止突出的颈椎盘的系统和方法。 用于警告前头姿势的系统包括：通过使用所获得的图像来测量用户的姿势的姿势确定装置，基于用户的姿势和预设的参考姿势来确定警告级别，并且提供所确定的警告级别 ; 以及警告生成装置，其生成与所述警告级别对应的警告，并向所述用户提供所述警告级别。 因此，根据用于警告前头姿势的系统和方法，在使用信息处理装置的环境中，通过在便携式外部设备中附加地使用警告功能来方便地识别前方头部姿势。 通过向用户发出关于姿势的警告，无论信息处理装置的位置和种类如何，都能有效地防止前方头部姿势症状和突出的颈椎盘。
-
公开(公告)号：KR1020140053504A
公开(公告)日：2014-05-08
申请号：KR1020120119508
申请日：2012-10-26
Applicant: 삼성전자주식회사 , 포항공과대학교 산학협력단
CPC classification number: G06K9/00288 , G06K9/00261 , G06K9/00268 , G06K9/6201 , G06K2009/4666
Abstract: A face recognition method according to an aspect of the present invention includes the following steps: an image obtaining step; an extraction step for extracting a first face image which is a facial area in the obtained image; a calculation step for calculating a feature value of the first image; a similarity calculation step for calculating the similarities between the first feature value and a second feature value of a second face image; a determination step for determining if the first and the second face images both belongs to the same user based on the results of the comparison between the similarities and the preset threshold value.
Abstract translation: 根据本发明的一个方面的面部识别方法包括以下步骤：图像获取步骤; 提取步骤，用于提取作为所获得图像中的面部区域的第一面部图像; 用于计算第一图像的特征值的计算步骤; 相似度计算步骤，用于计算第一特征值和第二面部图像的第二特征值之间的相似度; 基于相似度和预设阈值之间的比较结果，确定第一和第二面部图像是否都属于同一用户的确定步骤。
-
公开(公告)号：KR1020140000821A
公开(公告)日：2014-01-06
申请号：KR1020120068176
申请日：2012-06-25
Applicant: 한화테크윈 주식회사 , 포항공과대학교 산학협력단
CPC classification number: G06T7/20 , G06T5/005 , G06T2207/10048 , G06T2207/20081 , G06T2207/30196 , G06T2207/30232 , G06K9/00 , G06K9/00221 , G06K9/00335 , G06T5/00 , G06T7/143 , G06T7/35 , H04N5/30
Abstract: The present invention relates to a movement detecting system capable of removing thermal shadow, and a method thereof. According to an embodiment of the present invention, the movement detecting system capable of removing thermal shadow includes a database classifying data about the properties of the thermal shadow according to temperature and storing the same; a learning unit obtaining information on the size and brightness of the thermal shadow according to temperature, based on the properties of the thermal shadow; and a detecting unit removing the areas of the thermal shadow in a thermal image which is a target of detection. [Reference numerals] (310) Control unit; (320) Database; (330) Learning unit; (331) Thermal shadow size obtaining unit; (332) Thermal shadow brightness obtaining unit; (340) Detecting unit; (341) Candidate area designating unit; (342) Thermal shadow area designating unit; (343) Thermal shadow removing unit
Abstract translation: 本发明涉及能够去除热阴影的运动检测系统及其方法。 根据本发明的实施例，能够去除热阴影的移动检测系统包括：根据温度分类关于热影的属性的数据的数据库并存储; 学习单元，基于热阴影的性质，根据温度获得关于热阴影的尺寸和亮度的信息; 以及检测单元去除作为检测对象的热图像中的热阴影的区域。 （附图标记）（310）控制单元; （320）数据库; （330）学习单位; （331）热阴影尺寸获取单元; （332）热阴影亮度获取单元; （340）检测单元; （341）候选区指定单位; （342）热影区指定单位; （343）热阴影去除单元
-
公开(公告)号：KR100904916B1
公开(公告)日：2009-07-01
申请号：KR1020080057935
申请日：2008-06-19
Applicant: 주식회사 다우엑실리콘 , 주식회사 크라스아이디 , 포항공과대학교 산학협력단
CPC classification number: G06K9/00275 , G06K9/4633 , G06K9/4661 , G06K9/6202
Abstract: A system and a method for recognizing a face are provided to stably recognize a face in an environment in which various lightings are changed. A face extracting unit(130) extracts a face area from an input image as a face image. A conversion image computing unit(140) produces a conversion image by using a binary pattern by comparing each pixel of the face image with pixel values of adjacent pixels. A feature information extracting unit(150) extracts feature information about the face image by multiplying the conversion image by a projection matrix. A face recognizing unit(170) recognizes a face of the input image by comparing the extracted feature information with a character image stored in a database.
Abstract translation: 提供用于识别面部的系统和方法，以在各种照明改变的环境中稳定地识别脸部。 面部提取单元（130）从输入图像中提取面部区域作为面部图像。 转换图像计算单元（140）通过将面部图像的每个像素与相邻像素的像素值进行比较，通过使用二进制图案来生成转换图像。 特征信息提取单元（150）通过将转换图像乘以投影矩阵来提取关于脸部图像的特征信息。 面部识别单元（170）通过将所提取的特征信息与存储在数据库中的字符图像进行比较来识别输入图像的面部。
-
-
-
-
-
-
-
-
-
Search Results
64
records
(took 0.038 seconds)
