-
公开(公告)号：KR101330636B1
公开(公告)日：2013-11-18
申请号：KR1020070007663
申请日：2007-01-24
Applicant: 삼성전자주식회사
CPC classification number: G06K9/00288 , G06K9/4614 , G06K9/6257
Abstract: 얼굴시점 결정장치 및 방법과 이를 채용하는 얼굴검출장치 및 방법이 개시된다. 얼굴검출장치는 현재 영상이 얼굴에 해당하는지 여부를 판단하는 비얼굴 판단 부; 상기 현재 영상이 얼굴에 해당하는 경우 상기 현재 영상에 대하여 적어도 하나 이상의 시점 클래스를 추정하는 시점 추정부; 및 상기 추정된 적어도 하나 이상의 시점 클래스를 독립적으로 검증하여 상기 현재 영상의 최종적인 시점 클래스를 결정하는 독립 시점 검증부로 이루어진다.
-
公开(公告)号：KR1020120089000A
公开(公告)日：2012-08-09
申请号：KR1020110010035
申请日：2011-02-01
Applicant: 삼성전자주식회사
CPC classification number: G06F13/102 , G06F9/4415 , H04M1/72525 , H04M1/72533 , G06F8/61 , G06F9/445
Abstract: PURPOSE: An apparatus and a method for supplying an automatic application installing function are provided to increase coupling performance with a peripheral device and to supply a connection program from a digital device to a peripheral device. CONSTITUTION: A communication unit(112) receives system information from a peripheral device. The communication unit transmits a connection program list. A program confirming unit(104) analyzes the received system information. The program confirming unit generates a list of a connection program executed from the peripheral device. A control unit(100) transmits the generated connection program list to the peripheral device.
Abstract translation: 目的：提供一种用于提供自动应用安装功能的设备和方法，以增加与外围设备的耦合性能，并将连接程序从数字设备提供给外围设备。 构成：通信单元（112）从外围设备接收系统信息。 通信单元发送连接节目列表。 节目确认单元（104）分析所接收的系统信息。 程序确认单元生成从外围设备执行的连接程序的列表。 控制单元（100）将生成的连接节目列表发送到外围设备。
-
公开(公告)号：KR1020090082816A
公开(公告)日：2009-07-31
申请号：KR1020080008783
申请日：2008-01-28
CPC classification number: H04N5/232 , H04N13/246
Abstract: A multi-camera correcting method using a parallel laser beam and a system thereof are provided to enable calibration of a multi-camera even without a large-sized calibration grating required to obtain a pattern image, when a camera installation environment is large. A laser beam irradiator irradiates at least more than two pairs of visible laser beams(105) which are parallel to each other. Camera units(110,120,130) generate measurement values on a direction and a distance of the laser beams. Based on the measurement values and a relation of the direction and the distance between the laser beams, a camera variable extractor extracts directional values and parallel moving values among the cameras. A camera calibration unit calibrates the cameras based on the directional values and the parallel moving values.
Abstract translation: 提供了使用平行激光束的多摄像机校正方法及其系统，以便在摄像机安装环境较大时，即使没有获得图案图像所需的大尺寸校准光栅，也能够对多摄像机进行校准。 激光束照射器照射彼此平行的至少两对可见激光束（105）。 相机单元（110,120,130）在激光束的方向和距离上产生测量值。 基于测量值和激光束之间的方向和距离的关系，相机变量提取器提取相机中的方向值和平行移动值。 相机校准单元根据方向值和平行移动值校准摄像机。
-
公开(公告)号：KR1020090037718A
公开(公告)日：2009-04-16
申请号：KR1020070103198
申请日：2007-10-12
Applicant: 삼성전자주식회사
IPC: H04N7/01
CPC classification number: G06T3/40 , G06K9/00221 , G06T7/70
Abstract: An image converting method and an apparatus thereof are provided to restore an image of low resolution into an image of high resolution by using a resolution conversion matrix and an object model. An object region detector(110) detects an object area within a source image. Privacy is secured through low resolution processing including a mosaic process. The object region detector obtains both eye coordinates in the detected object region. A pause classifier(120) senses a pause of the detected object region. A resolution converting unit(150) converts the detected object region into low resolution by using a conversion matrix. An image mapping unit(160) maps the object region of low resolution on the source image. A storage unit receives a resolution conversion matrix. The storage unit receives and stores a generated object model.
Abstract translation: 提供一种图像转换方法及其装置，通过使用分辨率转换矩阵和对象模型，将低分辨率的图像恢复为高分辨率的图像。 对象区域检测器（110）检测源图像内的对象区域。 通过低分辨率处理（包括马赛克过程）来保护隐私。 对象区域检测器获得检测对象区域中的两个眼坐标。 暂停分类器（120）感测检测到的对象区域的暂停。 分辨率转换单元（150）通过使用转换矩阵将检测到的对象区域转换为低分辨率。 图像映射单元（160）在源图像上映射低分辨率的对象区域。 存储单元接收分辨率转换矩阵。 存储单元接收并存储生成的对象模型。
-
公开(公告)号：KR1020090036402A
公开(公告)日：2009-04-14
申请号：KR1020070101576
申请日：2007-10-09
Applicant: 삼성전자주식회사
IPC: H04N7/18
CPC classification number: H04N7/18 , G06T5/002 , G06T7/20 , G06T2207/20182 , G08B23/00
Abstract: A moving object detecting method and a system therefor are provided to obtain a bipolar difference image and divide noise from a moving object by using spatial distribution of two images. A difference image unit(100) generates a bipolar difference image by using a previous image and a current image. A distance value calculation unit(200) generates a supply node and a consumption node from a plus image and a minus image of the bipolar difference image. The distance value calculation unit generates a supply-consumption graph by using the supply node and the consumption node. A compensation value calculation unit(400) calculates a compensation value with regard to noise occurring between a previous image and a current image by using a calculated distance value. An object motion detection unit(300) detects movement of an object by using a compensation value about the calculated noise.
Abstract translation: 提供了一种移动物体检测方法及其系统，以通过使用两个图像的空间分布来获得双极差分图像并且从移动物体分离噪声。 差分图像单元（100）通过使用先前图像和当前图像来生成双极差分图像。 距离值计算单元（200）从双像差图像的正图像和负图像生成供给节点和消耗节点。 距离值计算单元通过使用供给节点和消耗节点来生成耗电量图。 补偿值计算单元（400）通过使用计算出的距离值来计算关于在先前图像和当前图像之间出现的噪声的补偿值。 物体运动检测单元（300）通过使用关于计算出的噪声的补偿值来检测物体的移动。
-
Patent Agency Ranking
公开(公告)号：KR100888476B1
公开(公告)日：2009-03-12
申请号：KR1020070016034
申请日：2007-02-15
Applicant: 삼성전자주식회사
CPC classification number: G06K9/00281 , G06K9/4619
Abstract: 본 발명은 얼굴이 포함된 영상에서 얼굴의 특징을 추출하는 방법 및 장치에 관한 것으로, 입력되는 영상을 영상 내의 소정의 위치들 각각에서 인식용 필터 세트로 필터링하고, 소정의 위치들 중에서 얼굴의 중앙을 기준으로 좌우가 대칭되는 위치들에서 필터링된 값들을 머징(Merging)한 후, 필터링된 값들과 머징된 값들을 합성함으로써, 얼굴의 특징을 추출하거나 비교함에 있어 수행되는 시간, 특징 값 및 저장 공간을 크게 줄일 수 있다. 또한, 낮은 하드웨어 사양에서도 잘 동작하는 얼굴 인식 시스템을 구현할 수 있다.
-
公开(公告)号：KR100860994B1
公开(公告)日：2008-09-30
申请号：KR1020070077176
申请日：2007-07-31
Applicant: 삼성전자주식회사
CPC classification number: H04N5/23212 , H04N5/23219 , H04N5/235
Abstract: A method and an apparatus for capturing an object-centered image are provided to check the position of an object and intensity of illumination in real time to capture an image with high picture quality and allow a user to photograph without controlling focus and exposure. An apparatus(100) for capturing an object-centered image includes an object detector(120), a controller(130) and a photographing unit(140). The object detector detects a previously registered object from an input image. The controller estimates photographing information on the detected object and generates control information for photographing the input image by using the estimated photographing information. The photographing unit photographs the input image according to the control information.
Abstract translation: 提供用于捕获物体中心图像的方法和装置，以实时检查物体的位置和照明强度，以捕捉具有高图像质量的图像，并允许用户拍摄而不控制对焦和曝光。 用于捕获物体中心图像的装置（100）包括物体检测器（120），控制器（130）和拍摄单元（140）。 物体检测器从输入图像检测先前登记的物体。 控制器估计检测对象上的拍摄信息，并且通过使用估计拍摄信息产生用于拍摄输入图像的控制信息。 拍摄单元根据控制信息拍摄输入图像。
-
公开(公告)号：KR1020080076293A
公开(公告)日：2008-08-20
申请号：KR1020070016034
申请日：2007-02-15
Applicant: 삼성전자주식회사
CPC classification number: G06K9/00281 , G06K9/4619
Abstract: A method and an apparatus for extracting features of a face from an image which contains the face are provided to reduce a time, a feature value and a storage space in extracting the features of the face or comparing the features. An apparatus for extracting features of a face includes the first filtering processor(400), the second filtering processor(410) and a merging unit(420). The first filtering processor receives a normalized image via an input terminal, receives plural pairs of Fiducila points which are symmetric with respect to a center of the face via another input terminal, and receives recognition filter sets via another input terminal. The second filtering processor receives a normalized image via another input terminal, receives plural pairs of Fiducila points which are not symmetric with respect to a center of the face via another input terminal, and receives recognition filter sets via another input terminal. The merging unit receives the first feature vectors extracted via the first filtering processor and merges symmetric components among the received first feature vectors.
Abstract translation: 提供一种从包含脸部的图像中提取脸部特征的方法和装置，用于在提取面部特征或比较特征时减少时间，特征值和存储空间。 用于提取面部特征的装置包括第一滤波处理器（400），第二滤波处理器（410）和合并单元（420）。 第一滤波处理器经由输入端子接收归一化图像，经由另一输入端子接收相对于面部中心对称的多对Fiducila点，并且经由另一输入端子接收识别滤波器组。 第二滤波处理器通过另一个输入端接收标准化图像，经由另一个输入端接收相对于面部中心不对称的多对Fiducila点，并通过另一输入端接收识别滤波器组。 合并单元接收经由第一滤波处理器提取的第一特征矢量并且合并所接收的第一特征向量中的对称分量。
-
49.发明公开얼굴 영상의 유사도 산출 방법 및 장치와 이를 이용한 얼굴영상 검색 방법 및 장치 그리고 얼굴 합성 방법 失效用于计算面部图像的相似性的方法和装置，用于检索脸部图像的方法和装置以及用于合成脸部图像的方法
公开(公告)号：KR1020080056591A
公开(公告)日：2008-06-23
申请号：KR1020060129680
申请日：2006-12-18
Applicant: 삼성전자주식회사
CPC classification number: G06K9/00268
Abstract: A method and a device for calculating similarity of a face image, the method and the device for retrieving the face image, and the method for synthesizing the face image are provided to improve reliability of face similarity and reduce complexity by reflecting global and local features of the face image on a similarity result. A global feature generator(40) generates global feature vector of a face image received through a receiver(10) by projecting the face image to a first basis for entire face area extracted from a training face image set. A local feature calculator(60) generates local feature vector of the inputted face image by projecting the inputted face image to a second basis for a local face area extracted from the training face image set. A final similarity calculator(80) calculates the similarity between a selected training face image and the inputted face image by using the global/local feature vectors according to one selected training face image and the inputted face image. A PCA(Principal Component Analysis) basis generator generates the first basis by performing PCA for the training face image set. An LFA(Local Feature Analysis) basis generator generates the second basis by performing LFA for the training face image set. A weight selection unit(70) bestows weight on the similarities calculated through a global feature calculator(50) and the local feature calculator.
Abstract translation: 提供一种用于计算面部图像的相似度的方法和装置，用于检索脸部图像的方法和装置以及用于合成面部图像的方法，以通过反映全局和局部特征来提高面部相似度的可靠性并降低复杂度 脸部图像的相似度结果。 全局特征生成器（40）通过对从训练面部图像集提取的整个面部区域的第一基础投影面部图像来生成通过接收器（10）接收的面部图像的全局特征向量。 局部特征计算器（60）通过将输入的面部图像投射到从训练面部图像集提取的局部面部区域的第二基础上来生成输入的面部图像的局部特征向量。 最终相似度计算器（80）根据所选择的训练面部图像和输入的面部图像，使用全局/局部特征向量来计算所选择的训练面部图像与输入的面部图像之间的相似度。 PCA（主成分分析）基础发生器通过执行训练面部图像集的PCA生成第一个基础。 基于LFA（局部特征分析）的发生器通过对训练面部图像集执行LFA来产生第二基础。 权重选择单元（70）赋予通过全局特征计算器（50）和局部特征计算器计算的相似度的权重。
-
公开(公告)号：KR1020080041931A
公开(公告)日：2008-05-14
申请号：KR1020060110170
申请日：2006-11-08
Applicant: 삼성전자주식회사
CPC classification number: G06K9/00288
Abstract: A method and an apparatus for recognizing a face by using extended Gabor wavelet features are provided to reduce errors in face recognition in accordance with illumination, facial expression or pose and to enhance recognition approval rate. A method for recognizing a face comprises the following several steps. The first feature extractor extends a Gabor wavelet filter(100). The first feature extractor applies the extended Gabor wavelet filter to a training face image which is resulted from a preprocessing procedure of a training face image preprocessor, and extracts Gabor wavelet features(200). A selector selects efficient Gabor wavelet features from the extracted Gabor wavelet features by using a boosting learning algorithm which is one among statistical resampling algorithms, and constructs a Gabor wavelet feature set(300). A linear discriminant analysis learning unit calculates a basis vector via a linear discriminant analysis(400). The second feature extractor extracts Gabor wavelet features from an inputted image by applying the Gabor wavelet set to the inputted image(500). A face descriptor generator generates face descriptors via projection with the basis vector by using the Gabor wavelet features extracted by the second feature extractor(600).
Abstract translation: 提供一种通过使用扩展Gabor小波特征来识别脸部的方法和装置，以减少根据照明，面部表情或姿势的脸部识别中的错误并提高识别批准率。 用于识别脸部的方法包括以下几个步骤。 第一个特征提取器扩展了Gabor小波滤波器（100）。 第一个特征提取器将扩展Gabor小波滤波器应用于由训练面部图像预处理器的预处理过程产生的训练面部图像，并提取Gabor小波特征（200）。 选择器通过使用作为统计重采样算法之一的增强学习算法，从提取的Gabor小波特征中选出有效的Gabor小波特征，并构建Gabor小波特征集（300）。 线性判别分析学习单元经由线性判别分析（400）计算基本向量。 第二特征提取器通过对输入的图像（500）应用Gabor小波集，从输入的图像中提取Gabor小波特征。 面部描述符生成器通过使用由第二特征提取器（600）提取的Gabor小波特征通过基准向量的投影生成面部描述符。
-
-
-
-
-
-
-
-
-
Search Results
69
records
(took 0.031 seconds)
