가버 특징과 SVM 분류기를 이용하여 위변조 얼굴을 인식하기 위한 시스템 및 그 방법

    公开(公告)号:KR101877981B1

    公开(公告)日:2018-07-12

    申请号:KR1020110139605

    申请日:2011-12-21

    CPC classification number: G06K9/00 G06K9/00288 G06K9/00899

    Abstract: 본발명에의한가버특징과 SVM 분류기를이용하여위변조얼굴을인식하기위한시스템및 그방법이개시된다. 본발명에따른위변조얼굴을인식하기위한시스템은다수의얼굴영상샘플들로부터하나의표준얼굴그래프를생성하는그래프생성수단; 다수의얼굴영상샘플과위변조얼굴영상샘플로부터위변조얼굴을판별하기위한최적분류평면을결정하는 SVM 학습수단; 및인식하고자하는얼굴영상을입력받으면, 상기표준얼굴그래프와상기최적분류평면을이용하여입력받은상기얼굴영상이위변조된얼굴인지를판별하는얼굴인식수단을포함한다. 본발명은주변조명영향이나얼굴의다양한패턴에대해서도효율적으로얼굴을인식할수 있을뿐 아니라, 얼굴인식의정확도를향상시킬수 있고, 얼굴인식에대한신뢰도를향상시킬수 있다.

    승마 시뮬레이터 및 승마 시뮬레이션 방법
    6.
    发明公开
    승마 시뮬레이터 및 승마 시뮬레이션 방법 审中-实审
    用于骑马的模拟器和用于模拟骑马的方法

    公开(公告)号:KR1020140043174A

    公开(公告)日:2014-04-08

    申请号:KR1020120108476

    申请日:2012-09-28

    Abstract: Disclosed are a simulator for horse riding which adaptively operates for users, and a method for simulating the horse riding. The simulator for horse riding comprises: a user identification part for identifying a user by using a user identification information according to a face of the user extracted from input images; a posture recognition part for outputting a user posture information by extracting a specific body part information of the user from the input images; a coaching part for providing the user an educational information about horse riding postures based on the user identification information and the user posture information; and an actual horseback riding sensing part for controlling the horse riding machine based on user intended postures which are output by analyzing the user posture information. Therefore, user-customized horse riding education can be provided through the identification of users and the recognition of the postures of users. In addition, an actual sensing type of simulation for horse riding can be provided by converting modes of the horse riding machine to correspond to the user intended postures. [Reference numerals] (100) User identification part; (110) Face detection unit; (120) Facial characteristic extraction unit; (200) Posture recognition part; (210) User region extracting unit; (220) Posture information outputting unit; (300) Coaching part; (310) Motion data storage part; (320) Model matching unit; (330) Horse riding education unit; (400) Actual horseback riding sensing part; (410) Intended posture outputting unit; (420) Machine controlling unit; (AA,BB) Input image

    Abstract translation: 公开了一种适用于用户的骑马模拟器,以及用于模拟骑马的方法。 用于骑马的模拟器包括:用户识别部分,用于根据从输入图像提取的用户的脸部使用用户识别信息来识别用户; 姿势识别部,其通过从所述输入图像提取所述用户的特定身体部位信息来输出用户姿势信息; 基于用户识别信息和用户姿势信息向用户提供关于骑马姿势的教育信息的辅导部分; 以及基于通过分析用户姿势信息输出的用户期望的姿势来控制骑马机的实际的骑马感测部。 因此,用户定制的骑马教育可以通过用户的识别和用户的姿态的识别来提供。 另外,可以通过将骑马机的模式转换为与用户期望的姿势对应来提供用于骑马的实际感测类型的模拟。 (附图标记)(100)用户识别部分; (110)面部检测单元; (120)面部特征提取单位; (200)姿态识别部分; (210)用户区域提取单元; (220)姿势信息输出单元; (300)教练部分; (310)运动数据存储部分; (320)型号匹配单元; (330)骑马教育单位; (400)实际骑马感应部分; (410)预期姿势输出单元; (420)机器控制单元; (AA,BB)输入图像

    화소 이동 문서 영상 조합 인식 방법을 이용한 카메라 문자인식 장치 및 방법
    7.
    发明公开
    화소 이동 문서 영상 조합 인식 방법을 이용한 카메라 문자인식 장치 및 방법 有权
    通过像素相机摄像机图像的组合识别方法识别信息的装置和方法

    公开(公告)号:KR1020080050272A

    公开(公告)日:2008-06-05

    申请号:KR1020070081866

    申请日:2007-08-14

    Abstract: An apparatus and a method for recognizing letters with a camera by using a pixel-shifted document image combined recognition method are provided to improve quality of images by using an image reinforcing algorithm and remove an unrecognized document area by executing partial binarizing, and improve performance of recognizing letters of a camera by extracting a characteristic considering shape deformation of letters and combining different recognizers. An apparatus for recognizing letters by using a camera includes a sub-document image processor(1) processing an input image into sub-document images, wherein pixels are shifted, a binarizing unit(2) partially binarizing the sub-document images passing through an image reinforcing algorithm, a letter processor(5,6) extracting structural characteristic information of an individual letter and a pixel mobile network characteristic of a letter area to which the pixels are shifted, and a recognition processor(3,4) combining different recognizers classifying letter types and recognizing the individual letters for outputting a result of recognition.

    Abstract translation: 提供一种通过使用像素偏移文档图像组合识别方法用相机识别字母的装置和方法,以通过使用图像增强算法来提高图像的质量,并且通过执行部分二值化来去除未被识别的文档区域,并且提高 通过提取考虑字母形状变形的特征并组合不同的识别器来识别相机的字母。 一种用于通过使用相机识别字母的装置包括:一个副图像处理器(1),将输入图像处理成其中像素被移位的子文档图像;二值化单元(2),将通过一个 图像增强算法,提取单个字母的结构特征信息的字母处理器(5,6)以及像素移动到的字母区域的像素移动网络特性;以及识别处理器(3,4),其组合不同的识别器分类 字母类型并识别用于输出识别结果的各个字母。

    기계 이상 동작 검출 장치 및 방법
    9.
    发明授权
    기계 이상 동작 검출 장치 및 방법 有权
    检测机器异常操作的设备和方法

    公开(公告)号:KR101830331B1

    公开(公告)日:2018-03-29

    申请号:KR1020160006082

    申请日:2016-01-18

    Abstract: 기계이상동작검출장치및 방법이개시된다. 본발명의일실시예에따른기계이상동작검출장치는촬영된기계의동작영상의주기성을분석하여검출된주기특징을동작패턴 DB에등록주기특징으로저장하는동작패턴등록부; 및상기등록주기특징과상기영상간의특징량을분석하여이상동작을검출하는이상동작검출부를포함한다.

    Abstract translation: 公开了一种用于检测机器的异常操作的设备和方法。 机器故障检测按照装置与本发明的一个实施例中,对于由拍摄机器周期,其特征在于由运动图形特征DB的图像分析存储运动的所检测的周期性的登记周期运动图案登记单元; 以及异常操作检测单元,用于分析注册周期特性和图像之间的特性以检测异常操作。

    포즈 변화에 강인한 얼굴 인식 시스템 및 방법
    10.
    发明公开
    포즈 변화에 강인한 얼굴 인식 시스템 및 방법 审中-实审
    面部识别系统和方法对姿势变化具有鲁棒性

    公开(公告)号:KR1020170140519A

    公开(公告)日:2017-12-21

    申请号:KR1020160072901

    申请日:2016-06-13

    Abstract: 본발명의일예와관련된얼굴인식시스템은대상사용자얼굴영상을획득하는영상획득부, 표준얼굴그래프를상기대상사용자얼굴영상에정합하고, 상기대상사용자얼굴영상에서상기표준얼굴그래프의노드가위치한지점의특징벡터와상기표준얼굴그래프의특징벡터의차이가기 설정된차이이내가되도록상기정합된표준얼굴그래프노드의위치를보정하여최적얼굴그래프를생성하는얼굴그래프생성부및 상기대상사용자얼굴영상에서상기최적얼굴그래프의노드가위치한지점별특징벡터와복수의사용자얼굴그래프의노드별특징벡터를비교하여상기대상사용자를인식하는사용자인식부를포함할수 있다.

    Abstract translation: 根据本发明示例性实施例的面部识别系统包括:图像获取单元,用于获取对象用户面部图像;面部图像匹配单元,用于将标准面部图形与目标用户面部图像进行匹配; 面部图形生成单元,用于通过校正所述匹配的标准面部图形节点的位置以使得所述标准面部图形的特征​​矢量和特征矢量之间的差异在预定的集合差值内来生成最优面部图形; 以及用户识别单元,用于通过比较图表的节点所处的每个点的特征向量和多个用户脸部图形的每个节点的特征向量来识别目标用户。

Patent Agency Ranking