-
公开(公告)号：KR1020110071207A
公开(公告)日：2011-06-29
申请号：KR1020090127711
申请日：2009-12-21
Applicant: 한국전자통신연구원
CPC classification number: H04N13/161 , H04N13/167 , H04N13/327
Abstract: PURPOSE: A camera image evaluation and correction method for stereo matching and a stereo camera system using the same are provided to offer right/left images for stereo matching which is properly controlled on the basis of the human. CONSTITUTION: A right and left motion vector separator(360) separates right and left motion vectors. A human detector(370) creates a left ROI(Region Of Interest) area and a right ROI area in a left image motion vector and a right image motion vector. An ROI representative value extractor(380) extracts a representative value in the left ROI image of the left ROI area and the right ROI image of the right ROI area.
Abstract translation: 目的：提供用于立体匹配的相机图像评估和校正方法以及使用其的立体相机系统，以提供基于人类适当控制的立体匹配的右/左图像。 构成：右和左运动矢量分离器（360）分离左右运动矢量。 人类检测器（370）在左图像运动矢量和右图像运动矢量中创建左ROI（感兴趣区域）区域和右ROI区域。 ROI代表值提取器（380）提取左ROI区域的左ROI图像和右ROI区域的右ROI图像中的代表值。
-
公开(公告)号：KR1020100124949A
公开(公告)日：2010-11-30
申请号：KR1020090043914
申请日：2009-05-20
Applicant: 한국전자통신연구원
IPC: H04N5/262
CPC classification number: G06K9/00255 , G06K9/00261 , G06T7/277 , G06T2207/30201
Abstract: PURPOSE: A device and a method for tracking a face at a long distance are provided to accurately detect and track a far-distanced very small face at a low resolution in order to prevent the wrong detection which tracks a non-facial portion. CONSTITUTION: An image input unit(110) generates continuous image frames by inputting a video image, and a face tracking initializing unit(130) detects an omega region and a facial region of a human from the image frames. The face tracking initializing unit sets the estimated facial region as tracking start region information. A remote face tracker(150) detects a facial region at a short distance within the facial detection region.
Abstract translation: 目的：提供用于跟踪远距离脸部的装置和方法，以便以低分辨率准确地检测和跟踪远距离的非常小的脸部，以防止跟踪非面部部分的错误检测。 构成：图像输入单元（110）通过输入视频图像来生成连续的图像帧，并且面部追踪初始化单元（130）从图像帧中检测人的Ω区域和面部区域。 面部跟踪初始化单元将估计的面部区域设置为跟踪开始区域信息。 远程脸部跟踪器（150）在脸部检测区域内检测距离较短距离的面部区域。
-
公开(公告)号：KR1020100102949A
公开(公告)日：2010-09-27
申请号：KR1020090021279
申请日：2009-03-12
Applicant: 한국전자통신연구원
IPC: G06T7/40
CPC classification number: G06K9/00275 , G06K9/6234 , G06K9/6247 , G06T7/11 , G06T2207/20021
Abstract: PURPOSE: A method and an apparatus for recognizing a face by a block unit are provided to determine similarity by comparing face images by a block unit, thereby producing various combinations of images to increase variety of face images. CONSTITUTION: A query face image is divided into fixed blocks and a feature value is extracted from each block. A feature value extracting unit(200) extracts a feature value from an inputted face image by a fixed block unit. A block similarity calculating unit(410) calculates block similarity between the query face image and a registration face image. The block similarity calculating unit compares a feature value of a fixed block of the query face image with a feature value of a block of the registration image corresponding to a fixed block of the query face image.
Abstract translation: 目的：提供一种用于通过块单元识别面部的方法和装置，以通过以块为单位比较面部图像来确定相似度，从而产生图像的各种组合以增加脸部图像的多样性。 构成：将查询面部图像分为固定块，并从每个块中提取特征值。 特征值提取单元（200）通过固定块单元从输入的面部图像提取特征值。 块相似度计算单元（410）计算查询面部图像与登记面部图像之间的块相似度。 块相似度计算单元将查询面部图像的固定块的特征值与对应于查询面部图像的固定块的登记图像的块的特征值进行比较。
-
公开(公告)号：KR1020100072772A
公开(公告)日：2010-07-01
申请号：KR1020080131279
申请日：2008-12-22
Applicant: 한국전자통신연구원
CPC classification number: G06K9/4642 , G06K9/00228
Abstract: PURPOSE: A method and an apparatus for real-time face detection using a stereo vision are provided to perform face detection in foreground including face pattern from whole video by stereo vision processor thereby widely reducing face detection processing time. CONSTITUTION: A vision processing unit(320) calculates distance information by usage of a plurality of images including face pattern. Based on the distance information, the vision processing unit divides foreground image including the face pattern and background image not including the face pattern. According to distance information, a face detection unit(330) adjusts video size of the foreground image. The face detection unit detects the face pattern from adjusted foreground image.
Abstract translation: 目的：提供一种使用立体视觉进行实时脸部检测的方法和装置，用于在前景执行面部检测，包括立体视觉处理器的整个视频的脸部图案，从而大大减少脸部检测处理时间。 构成：视觉处理单元（320）通过使用包括脸部图案的多个图像来计算距离信息。 基于距离信息，视觉处理单元将包括脸部图案的前景图像和不包括脸部图案的背景图像分割开。 根据距离信息，面部检测单元（330）调整前景图像的视频大小。 面部检测单元从调整前景图像检测脸部图案。
-
公开(公告)号：KR1020100055168A
公开(公告)日：2010-05-26
申请号：KR1020080114116
申请日：2008-11-17
Applicant: 한국전자통신연구원
Abstract: PURPOSE: A speaker identification method for a multi-channel speaker identification system is provided to improve the performance of speaker identification and speaker verification in a remote distance multi-channel environment. CONSTITUTION: A speaker identifier for a multi-channel speaker identification system is connected to each input microphone(101). A speaker identification integrator(115) integrates the identification result of the speaker identifier. The channel speaker identifier computes the degree of speaker identification contribution by the measurement of posterior probability of a registered speaker. The speaker identification integrator produces channel reliability through speaker identification contribution. The speaker identification integrator integrates the channel speaker identification result with the weight application by the channel reliability.
Abstract translation: 目的：提供多声道扬声器识别系统的扬声器识别方法，以提高远程多声道环境中扬声器识别和扬声器验证的性能。 构成：用于多声道扬声器识别系统的扬声器标识符连接到每个输入麦克风（101）。 扬声器识别积分器（115）集成了扬声器标识符的识别结果。 信道扬声器标识符通过测量注册说话者的后验概率来计算扬声器识别贡献的程度。 扬声器识别积分器通过扬声器识别贡献产生通道可靠性。 扬声器识别积分器通过信道可靠性将声道扬声器识别结果与重量应用相结合。
-
Patent Agency Ranking
公开(公告)号：KR100951637B1
公开(公告)日：2010-04-09
申请号：KR1020070132691
申请日：2007-12-17
Applicant: 한국전자통신연구원
Abstract: 본 발명은 스테레오 비전 시스템의 영상 처리 방법에 관한 것으로, 이를 위하여 본 발명은, 스테레오 영상 입력에 대한 전처리 과정을 수행하고, 동일한 물체로 판단되는 좌측 및 우측 영상의 픽셀간 양안차(disparity)를 생성하는 스테레오 매칭을 수행하여 스테레오 영상 정보를 획득하는 종래 방법과는 달리, 스테레오 비전 시스템에서 스테레오 영상이 입력되면, 영상 압축 모듈을 통해 저해상도 거리 정보를 획득하거나 스테레오 매칭 모듈을 이용하여 고해상도 거리 정보를 획득한 후, 저해상도 거리 정보 또는 고해상도 거리 정보에 대응하는 영상의 움직임 벡터를 이용하여 영상 노이즈를 제거함으로써, 움직임 벡터를 이용하여 영상 노이즈를 효과적으로 제거하여 스테레오 영상에 대한 정확한 거리 정보를 획득할 수 있는 것이다.
스테레오 비전 시스템, 스테레오 매칭(Stereo Matching)-
公开(公告)号：KR1020100015167A
公开(公告)日：2010-02-12
申请号：KR1020080076098
申请日：2008-08-04
Applicant: 한국전자통신연구원
CPC classification number: G01H17/00 , G06K9/00348 , G06K9/0053
Abstract: PURPOSE: A user recognition device and a user recognition method are provided to use a walking pattern as ID information of a user, thereby improving a recognition rate of the user through simple configuration. CONSTITUTION: A sensing unit(110) senses vibration due to walking of a user to output an electrical signal. A pattern calculator(130) obtains a walking pattern of a pedestrian from the electrical signal. A user decider(150) compares the walking pattern with reference date. The user decider identifies a user based on the comparison result.
Abstract translation: 目的：提供用户识别装置和用户识别方法以使用步行模式作为用户的ID信息，从而通过简单的配置提高用户的识别率。 构成：感测单元（110）感测由于用户的行走而产生的振动以输出电信号。 图案计算器（130）从电信号中获取行人的步行图案。 用户决定器（150）将步行模式与参考日期进行比较。 用户决定器基于比较结果识别用户。
-
公开(公告)号：KR1020090065214A
公开(公告)日：2009-06-22
申请号：KR1020070132691
申请日：2007-12-17
Applicant: 한국전자통신연구원
CPC classification number: H04N13/128 , H04N19/13 , H04N19/57 , H04N2013/0081 , H04N2013/0085
Abstract: A stereo vision system and an image processing method thereof are provided to efficiently remove image noise by using a motion vector of an image, thereby obtaining exact distance information on a stereo image. An inputted stereo image is pre-processed. Low-resolution distance information using an image compression module about the pre-processed stereo image is obtained(616). High-resolution distance information is obtained through a stereo matching module(622). A motion vector is extracted by utilizing the image compression module. Image noise is eliminated by using the high-resolution distance information and the extracted motion vector(624). The motion vector is extracted through a motion estimation technique by alternately inputting the stereo image.
Abstract translation: 提供立体视觉系统及其图像处理方法，以通过使用图像的运动矢量来有效地去除图像噪声，从而获得关于立体图像的精确距离信息。 输入的立体声图像被预处理。 获得使用关于预处理立体图像的图像压缩模块的低分辨率距离信息（616）。 通过立体匹配模块（622）获得高分辨率距离信息。 利用图像压缩模块提取运动矢量。 通过使用高分辨率距离信息和提取的运动矢量来消除图像噪声（624）。 通过交替输入立体图像通过运动估计技术提取运动矢量。
-
公开(公告)号：KR100902343B1
公开(公告)日：2009-06-12
申请号：KR1020070113682
申请日：2007-11-08
Applicant: 한국전자통신연구원
IPC: G05D1/02
CPC classification number: H04N7/18 , G06K9/00664 , G06T7/593 , G06T2207/10021 , G06T2207/20021 , G06T2207/20052 , G06T2207/30196 , G06T2207/30261
Abstract: 본 발명은 로봇 영상 시스템 및 검출 방법에 관한 것으로, 이동형 로봇이 사람의 위치를 검출하고, 움직이는 사람의 방향으로 장애물을 피해 접근하도록 하는 서비스를 제공함으로써, 이동형 로봇의 주의를 끌고자 손을 흔들거나 로봇 앞에서 움직이는 사람을 찾는 일을 비교적 적은 연산만으로도 수행하여 메인 프로세서와 배터리를 적게 소모하므로 이동형 로봇의 대기시간을 향상시킬 수 있다.
이동형 로봇, 영상, 사람, 장애물Abstract translation: 用于输出视差图的机器人视觉系统包括用于接收左图像和右图像并在两个图像之间输出视差图的立体相机; 用于将左图像或右图像编码为基于运动补偿的视频比特流的编码器; 以及用于从视频比特流中提取图像块的编码类型，运动矢量和DCT系数的解码器。 此外，该系统包括人物检测器，用于使用左图像和右图像之间的视差图，块编码类型和运动矢量来检测和标记图像中的人物块，并且检测从标记的人到 相机; 以及用于检测比使用从视频比特流提取的块编码类型，运动矢量和DCT系数的人和视差图更接近的障碍物的障碍物检测器。
-
公开(公告)号：KR100783134B1
公开(公告)日：2007-12-07
申请号：KR1020060055956
申请日：2006-06-21
Applicant: 한국전자통신연구원
IPC: B25J5/00
Abstract: 본 발명은 로봇의 무게중심 이동장치, 그 장치를 이용한 시스템 및 방법에 관한 것으로, 로봇의 본체 내부에 소정의 무게를 가진 이동질량을 일정한 방향으로 이송하여 로봇의 무게중심을 이동하는 질량이송장치를 구비하여 로봇이 주행하는 과정에서 구동력이 없는 캐스터가 장애물에 의해 더 이상 구동할 수 없는 경우에 장애물을 넘어서 계속 주행할 수 있는 효과가 있다.
이동질량, 무게중심, 질량이송장치
-
-
-
-
-
-
-
-
-
Search Results
170
records
(took 0.032 seconds)
