-
-
公开(公告)号：KR101905426B1
公开(公告)日：2018-10-10
申请号：KR1020120003295
申请日：2012-01-11
Applicant: 한국전자통신연구원
CPC classification number: G06K9/00865 , G06K9/3258 , G06K9/348 , G06K2209/01
Abstract: 본발명은문자인식의정확도를높이기위하여문자정보를순서화하는방법에관한것으로, 영상으로부터문자정보를추출하여추출된문자정보를미리결정된임계값을통하여이진화(binarization)하고, 이진화된문자정보의중심선을추출하여세선화(thinning)하는전처리하고, 전처리된문자정보를미리설정된기준에따른문자정보로정규화(nomailzation)한다음정규화된문자정보를끝점또는분기점을포함하는구조적특징을이용하여순서화하는단계를포함한다. 본발명에따르면입력문자정보의각도정규화방법, 구조적특징위치결정방법, 구조적특징숫자열생성방법등을제안함으로써, 자연영상으로부터얻은다양한폰트의문자를각도및 크기와무관하게강인하게인식할수 있다.
-
公开(公告)号：KR1020170084639A
公开(公告)日：2017-07-20
申请号：KR1020160003911
申请日：2016-01-12
Applicant: 한국전자통신연구원
Abstract: 다중센서기반층간소음알림장치및 방법이개시된다. 본발명에일실시예에따른다중센서기반층간소음알림장치는소음발생대상의발목및 발바닥에설치된복수개의센서들로부터감지된신호에기반하여소음발생행위정보를생성하는센서부;경보를발생하는경보부및 상기소음발생행위정보에기반하여상기경보부를제어하는제어부를포함한다.
Abstract translation: 公开了一种基于多传感器的层间噪声通知装置和方法。 一个实施例中是按照从多个脚踝和通告击打噪声受试者，以产生本发明中的噪音行为信息传感器部的脚传感器的检测到的信号的基础上，基于传感器的层间噪声的示例;报警单元，用于产生一个报警 以及用于基于噪音生成活动信息来控制警报单元的控制器。
-
公开(公告)号：KR1020170082074A
公开(公告)日：2017-07-13
申请号：KR1020160001244
申请日：2016-01-05
Applicant: 한국전자통신연구원
CPC classification number: G06K9/00275 , G06K9/00248 , G06K9/00281 , G06K9/00288 , G06K9/6215
Abstract: 얼굴인식의정확도를높인관상학적특징정보를이용한얼굴인식기술이개시된다. 이를위해, 본발명의일 실시예에따른관상학적특징정보를이용한얼굴인식방법은얼굴구성요소별로관상학적표준타입들을정의하는단계; 사용자의얼굴영상을촬영하는단계; 상기얼굴영상에서얼굴구성요소정보를검출하는단계; 및상기얼굴구성요소정보에기초하여, 상기사용자의얼굴구성요소별로관상학적표준타입들과의유사도스코어를산정하는단계를포함하는것을특징으로한다.
Abstract translation: 公开了一种使用增加人脸识别准确性的冠状特征信息的人脸识别技术。 为此，根据本发明实施例的使用冠状特征信息来识别面部的方法包括：为每个面部部件定义冠状标准类型; 捕捉用户的脸部图像; 从脸部图像检测脸部分量信息; 并且基于脸部分量信息计算针对用户的每个脸部分量的相对于冠状标准类型的相似性分数。
-
公开(公告)号：KR101653235B1
公开(公告)日：2016-09-12
申请号：KR1020160029869
申请日：2016-03-11
Applicant: 한국전자통신연구원
Abstract: 제스쳐인식장치가개시된다. 이제스쳐인식장치는입력영상으로부터사용자의얼굴영역을검출하는휴먼검출부와, 상기검출된얼굴영역을기준으로상기사용자의팔의제스쳐가발생하는제스쳐영역을설정하는제스쳐영역설정부와, 상기제스쳐영역내에존재하는상기사용자의팔 영역을검출하는팔 검출부및 상기제스쳐영역내에존재하는팔 영역의위치, 이동방향성및 형태정보를분석하여, 상기사용자의목표제스쳐를판별하는제스쳐판정부를포함한다. 이러한제스쳐인식장치에의하면, 로봇이사용자의음성을인식하기어려운원거리에서인간로봇상호작용을위한유용한수단으로활용될수 있다.
-
Patent Agency Ranking
公开(公告)号：KR1020140100353A
公开(公告)日：2014-08-14
申请号：KR1020130013562
申请日：2013-02-06
Applicant: 한국전자통신연구원
IPC: G06T7/00
CPC classification number: H04N7/18 , G06K9/00335 , G06K9/00369 , G06K9/00664
Abstract: The present invention relates to a method and a device for recognizing human information. The disclosed method for recognizing human information comprises a step of generating sensor data-based human information comprising the identification, the location, and the behavior information of a human body existing in a recognition space by analyzing sensor data from multiple sensor resources arranged in the recognition space; a step of generating fused human information by fusing human information from a robot and human information from the sensor data, obtained and provided through interaction between a mobile robot terminal located in the recognition space and a human located in the recognition space, according to the location of the mobile robot and the state of the interaction; and a step of storing a human module according to the fused human information with respect to the human existing in the recognition space, in a database. Accordingly, the present invention is advantageous in improving the reliability of recognition information on the identification, the location, and the behavior information of a user when a plurality of users exist, by fusing multiple sensor resources installed in the recognition space and resources from a robot.
Abstract translation: 本发明涉及用于识别人类信息的方法和装置。 所公开的用于识别人类信息的方法包括通过分析来自布置在识别中的多个传感器资源的传感器数据，生成基于传感器数据的人类信息的步骤，该信息包括存在于识别空间中的人体的识别，位置和行为信息 空间; 根据所述位置，通过将位于所述识别空间中的移动机器人终端与位于所述识别空间中的人的位置之间的交互获得并提供的所述传感器数据，通过从所述机器人融合人类信息和人类信息来生成融合的人类信息的步骤; 的移动机器人和相互作用的状态; 以及在数据库中存储相对于存在于识别空间中的人的融合人类信息的人类模块的步骤。 因此，本发明在存在多个用户时通过融合安装在识别空间中的多个传感器资源和来自机器人的资源来提高对用户的识别，位置和行为信息的识别信息的可靠性是有利的 。
-
公开(公告)号：KR1020140049157A
公开(公告)日：2014-04-25
申请号：KR1020120114608
申请日：2012-10-16
Applicant: 한국전자통신연구원
IPC: G06Q50/10 , H04N21/4223
CPC classification number: H04N21/4755 , G06K9/00221 , G06K9/00335 , G06K2009/00322 , G06Q50/10
Abstract: Disclosed is a technology for providing a personalized service through behavioral analysis on detected behaviors of a user. A method for providing a personalized service includes a step of receiving input images including images of a user and calculating the location of the user by tracking the location of the user from the inputted images; a step of calculating facial data and posture data of the user based on the location of the user, performing behavioral analysis on the user using the facial data and posture data of the user, and calculating user behavioral analysis data including data about services which the user prefers; a step of updating user data using the location and behavioral analysis data of the user; and a step of determining a personalized service based on the updated user data. In particular, it is possible to provide programs or advertisements which a viewer prefers by detecting a viewer who watches a TV and analyzing behaviors of the viewer. [Reference numerals] (AA) Start; (BB) End; (S100) Receive an input image; (S200) Recognize a user; (S300) Analyze user behaviors; (S400) Update a user database; (S500) Select a preferred service based on user data; (S600) Provide the selected service
Abstract translation: 公开了一种通过针对用户的检测到的行为的行为分析来提供个性化服务的技术。 一种用于提供个性化服务的方法包括：接收包括用户图像的输入图像并通过从所输入的图像中跟踪用户的位置来计算用户的位置的步骤; 基于用户的位置来计算用户的面部数据和姿势数据的步骤，使用用户的面部数据和姿势数据对用户进行行为分析，以及计算用户行为分析数据，其包括关于用户的服务的数据 喜欢; 使用用户的位置和行为分析数据更新用户数据的步骤; 以及基于更新的用户数据确定个性化服务的步骤。 特别地，可以通过检测观看电视的观众和分析观众的行为来提供观众喜欢的节目或广告。 （附图标记）（AA）开始; （BB）结束; （S100）接收输入图像; （S200）识别用户; （S300）分析用户行为; （S400）更新用户数据库; （S500）根据用户数据选择首选服务; （S600）提供所选服务
-
公开(公告)号：KR1020140049152A
公开(公告)日：2014-04-25
申请号：KR1020120114537
申请日：2012-10-16
Applicant: 한국전자통신연구원
IPC: G05D1/12
CPC classification number: G05D1/0231 , G05D1/0248 , G06K9/00342 , G06T7/20 , G06T2207/10024 , G06T2207/10028 , G06T2207/30196
Abstract: Disclosed are a method for following a person and a robot apparatus. In the method for following a person, an image frame consisting of a color image and a depth image is provided. It is determined whether user flowing succeeds in a previous image frame. If user flowing succeeds in a previous image frame, the target point of a user position and device is determined based on the color image and the depth image in the image frame. The method for following a person predicts the present position of the user from the depth image, quickly flows the user, and flows the user by quickly redetecting the user by using the information of the user which is obtained in a flowing process when detecting the user fails due to an obstacle and so on. [Reference numerals] (AA) Image frame input; (BB,EE,FF,II) NO; (CC,DD,GG,JJ) YES; (HH) Next image frame; (S100) User flowing succeeds in a previous image frame?; (S111) Perform depth filtering; (S113) Perform following based on colors; (S115) Is the user flowing failed?; (S117) Maintaining an existing movement target point to set a following failure; (S119) Setting a user position as the movement target point; (S131) Detecting a head and shoulders; (S133) Perform comparison with the user; (S135) Is the user detection succeeded?; (S137) Setting the following succeed setting user position as the movement target point; (S139) Maintaining the existing target point; (S151) Start to move to the target point; (S153) Is the front obstacle confirmed?; (S155) Movement to the target; (S157) Movement to the target by avoiding the obtacle
Abstract translation: 公开了一种跟随人和机器人装置的方法。 在跟随人的方法中，提供了由彩色图像和深度图像组成的图像帧。 确定用户流动是否在先前的图像帧中成功。 如果用户流动在先前图像帧中成功，则基于图像帧中的彩色图像和深度图像来确定用户位置和设备的目标点。 用于跟随人的方法从深度图像预测用户的当前位置，快速地流过用户，并且通过使用在检测到用户时在流程中获得的用户的信息快速重新检测用户来流动用户 由于障碍等而失败。 （标号）（AA）图像帧输入; （BB，EE，FF，II）NO; （CC，DD，GG，JJ）是; （HH）下一帧; （S100）用户流动在先前的图像帧中成功？ （S111）进行深度滤波; （S113）基于颜色执行以下操作： （S115）用户是否流动失败？ （S117）维持现有的移动目标点设定以下故障; （S119）将用户位置设置为移动目标点; （S131）检测头肩; （S133）与用户进行比较; （S135）用户检测是否成功？ （S137）将用户位置设定为移动目标点以下， （S139）维持现有目标点; （S151）开始移动到目标点; （S153）前面的障碍确认了吗？ （S155）向目标移动; （S157）通过避免敌对行动向目标移动
-
公开(公告)号：KR1020140036803A
公开(公告)日：2014-03-26
申请号：KR1020120103375
申请日：2012-09-18
Applicant: 한국전자통신연구원
CPC classification number: A61B5/1176 , G06K9/00221 , G06K9/00906
Abstract: The present invention relates to a device and a method for distinguishing a disguised face by extracting a skin color area from a face area and confirming existence of a pulse element in the extracted skin color area. The device for distinguishing a disguised face according to an embodiment of the present invention includes a face area detecting unit detecting the face area in an input image inputted from the outside; a skin color modeling unit detecting the skin color area in the face area; and a disguised face distinguishing unit distinguishing whether a face in the input image is the disguised face or not by confirming whether the pulse element exists in a signal of the skin color area. [Reference numerals] (110) Face area detecting unit; (122) Converting unit; (124) Threshold calculating unit; (126) Modeling unit; (130) Signal calculating unit; (135) Buffer; (140) Signal processing unit; (150) Disguised face distinguishing unit
Abstract translation: 本发明涉及通过从面部区域提取皮肤颜色区域并确认提取的肤色区域中的脉冲元素的存在来区分伪装脸部的装置和方法。 根据本发明的实施例的用于区分伪装的面部的装置包括：面部检测单元，其检测从外部输入的输入图像中的面部区域; 皮肤颜色建模单元，其检测面部区域中的肤色区域; 以及伪装的面部识别单元，通过确认脉冲元素是否存在于肤色区域的信号中来区分输入图像中的脸部是否是伪装的脸部。 （附图标记）（110）面部区域检测单元; （122）转换单元; （124）阈值计算单元; （126）建模单元; （130）信号计算单元; （135）缓冲液 （140）信号处理单元; （150）伪装脸部识别单元
-
公开(公告)号：KR1020130072058A
公开(公告)日：2013-07-01
申请号：KR1020110139605
申请日：2011-12-21
Applicant: 한국전자통신연구원
CPC classification number: G06K9/00 , G06K9/00288 , G06K9/00899
Abstract: PURPOSE: A system for recognizing a forged and altered face using gabor feature and a support vector machine (SVM) classification machine and a method thereof are provided to recognize the face efficiently as to peripheral lighting influence or various patterns of the face, by estimating feature point positions of the face from the face area and extracting the gabor feature from the estimated face feature points and using the gabor feature as an input vector of the SVM classification machine. CONSTITUTION: A graph generation unit (100) generates a standard face graph from face image samples. A support vector machine (SVM) learning unit (200) determines an optimum classification plane to discriminate a forged and altered face from face image samples and forged and altered face image samples. A face recognition unit (300) determines forgery of an inputted face image, using the optimum classification plane. The face recognition unit detects a rectangular face area from the face image, and normalizes the rectangular face area into a fixed size, and generates an optimum face graph, using adaboost algorithm. [Reference numerals] (100) Graph generation unit; (200) SVM learning unit; (300) Face recognition unit
Abstract translation: 目的：提供一种使用陀飞轮特征识别伪造和改变脸部的系统，并提供支持向量机（SVM）分类机及其方法，通过估计特征来识别面部周边照明影响或脸部各种图案的面部 从脸部区域的脸部的点位置，并从估计的面部特征点提取出gabor特征，并使用gabor特征作为SVM分类机器的输入向量。 构成：图形生成单元（100）从面部图像样本生成标准面图。 支持向量机（SVM）学习单元（200）确定最佳分类平面以区分伪造和改变的面部与面部图像样本和伪造和改变的面部图像样本。 面部识别单元（300）使用最佳分类平面来确定所输入的面部图像的伪造。 面部识别单元从面部图像检测矩形脸部区域，并将矩形面部区域规范化为固定尺寸，并使用adaboost算法生成最佳面部图形。 （附图标记）（100）图形生成单元; （200）SVM学习单元; （300）脸部识别单元
-
-
-
-
-
-
-
-
-
Search Results
130
records
(took 0.036 seconds)
