-
公开(公告)号：KR101035291B1
公开(公告)日：2011-05-18
申请号：KR1020090126823
申请日：2009-12-18
Applicant: 전남대학교산학협력단
CPC classification number: G06F17/30259 , G06F17/16 , G06T7/223
Abstract: PURPOSE: A pose recognizing system using a joint model database is provided to improve a recognition speed and recognition rate. CONSTITUTION: A preprocessor(120) obtains a foreground image from an input image and changes the foreground into a distance converting image. A key pose recognizer(130) generates a joint model and a key pose eigenvectors and stores a pose library(110). The key pose recognizer extracts a key pose characteristic vector which is the most similar with a current image.
Abstract translation: 目的：提供使用联合模型数据库的姿态识别系统，以提高识别速度和识别率。 构成：预处理器（120）从输入图像获得前景图像并将前景改变为距离转换图像。 关键姿势识别器（130）产生联合模型和关键姿势特征向量并存储姿势库（110）。 关键姿态识别器提取与当前图像最相似的关键姿势特征向量。
-
公开(公告)号：KR1020110032429A
公开(公告)日：2011-03-30
申请号：KR1020090089906
申请日：2009-09-23
Applicant: 삼성전자주식회사 , 전남대학교산학협력단
CPC classification number: H04N5/262 , G06F3/017 , G06F17/30286 , H04B1/3827 , H04B1/401
Abstract: PURPOSE: An apparatus for improving gesture recognizing ratio in a mobile device is provided to generate integrated specification information from a plurality of input data, and recognize a gesture. CONSTITUTION: A gesture recognition unit(102) comprises a data acquisition unit(104), a feature extraction unit(106), and an integrated information generator(108). The data acquisition unit acquires a plurality of input data through cameras. The feature extraction unit performs pre-processing about a plurality of input data. The feature information extraction unit classifies the domain capable of gesture recognition. The integrated information generator mixes the extracted feature information to one integrated feature information.
Abstract translation: 目的：提供一种用于改善移动设备中的手势识别比率的装置，用于从多个输入数据生成综合指定信息，并识别手势。 构成：手势识别单元（102）包括数据获取单元（104），特征提取单元（106）和综合信息生成器（108）。 数据采集单元通过照相机获取多个输入数据。 特征提取单元执行关于多个输入数据的预处理。 特征信息提取单元对能够进行手势识别的域进行分类。 综合信息发生器将所提取的特征信息与一个综合特征信息进行混合。
-
公开(公告)号：KR1020090119442A
公开(公告)日：2009-11-19
申请号：KR1020080045490
申请日：2008-05-16
Applicant: 전남대학교산학협력단
IPC: H04N13/02
Abstract: PURPOSE: A recognition method of three dimensional human pose using a single camera is provided to estimate the photographing direction of human body by comparing angle histogram obtained from two dimensional pose image. CONSTITUTION: A three dimensional human body pose method for recognizing by a single camera is as follows. A target object having arbitrary three dimensional pose is photographed to obtain standard two dimensional pose images(S1000). A silhouette image of each standard two dimensional pose image is extracted, and the standard pose model is created based on the standard silhouette image. The standard pose model is stored in a standard pose model database. Standard pose models are created after repeating the first or third step, and stores in standard pose model database. The detection two dimensional pose image which takes a photograph of the detection object in the arbitrary direction by the single camera is gotten. The detected two dimensional pose image detection silhouette image is extracted which is based on the detected silhouette image and the video feature of the detection object. The standard pose model is obtained which has a closest similarity with the detection pose model.
Abstract translation: 目的：提供使用单个摄像机的三维人体姿态识别方法，通过比较从二维姿态图像获得的角度直方图来估计人体摄影方向。 构成：用于单个摄像机识别的三维人体姿态方法如下。 拍摄具有任意三维姿态的目标物体以获得标准二维姿态图像（S1000）。 提取每个标准二维姿态图像的剪影图像，并根据标准轮廓图像创建标准姿态模型。 标准姿态模型存储在标准姿态模型数据库中。 在重复第一步或第三步之后创建标准姿态模型，并存储在标准姿态模型数据库中。 得到了通过单个摄像机在任意方向拍摄检测对象的检测二维姿态图像。 提取检测到的二维姿态图像检测轮廓图像，其基于检测到的轮廓图像和检测对象的视频特征。 获得与检测姿态模型最相似的标准姿态模型。
-
公开(公告)号：KR101799270B1
公开(公告)日：2017-11-21
申请号：KR1020100137503
申请日：2010-12-29
Applicant: 엘지전자 주식회사 , 전남대학교산학협력단
CPC classification number: G06F3/017 , G06F3/0425 , G06F3/04883 , G06F2203/04808
Abstract: 본발명은, 터치스크린을이용하여복수의터치를동시에입력받는경우상기입력된복수의터치를포함하는복수의영상을일정시간동안연속입력받고, 상기복수의터치각각의터치패턴및 터치패턴변화를상기복수의영상각각에대하여판단하고, 상기판단결과에따라상기복수의터치에의한제스처를인식하는이동단말기및 이것의터치인식방법에관한것이다.
Abstract translation: 本发明涉及一种触摸输入装置和方法，当使用触摸屏同时输入多个触摸时，所述触摸输入装置和方法在预定时间连续地输入包括多个输入触摸的多个图像， 移动终端及其触摸识别方法技术领域本发明涉及根据确定结果通过多次敲击来判断多个图像并识别手势的移动终端及其触摸识别方法。
-
公开(公告)号：KR101165335B1
公开(公告)日：2012-07-18
申请号：KR1020110036082
申请日：2011-04-19
Applicant: 전남대학교산학협력단
CPC classification number: G06T7/277 , G06T2207/10016
Abstract: PURPOSE: A particle resampling method based on an index function and an image object tracking method using the same are provided to enable a user to effectively track an image object according to a purpose. CONSTITUTION: An image object is determined wherein the image object is a tracking target. Particles of a particle filter are initialized(S1000). A location of the image object is determined in a current location(S2000). Prediction particles predicted for determining the image object are generated by initial particles on which prediction particles of a next image are calculated based(S3000). It is determined whether an object determined as the image object is located in the current image(S4000).
Abstract translation: 目的：提供基于索引函数的粒子重采样方法和使用其的图像对象跟踪方法，以使用户能够根据目的有效地跟踪图像对象。 构成：确定其中图像对象是跟踪目标的图像对象。 粒子滤波器的粒子被初始化（S1000）。 在当前位置确定图像对象的位置（S2000）。 预测用于确定图像对象的预测粒子由其上计算下一图像的预测粒子的初始粒子产生（S3000）。 确定确定为图像对象的对象是否位于当前图像中（S4000）。
-
Patent Agency Ranking
公开(公告)号：KR1020120045627A
公开(公告)日：2012-05-09
申请号：KR1020100107284
申请日：2010-10-29
Applicant: 전남대학교산학협력단
CPC classification number: G06F3/017 , G06F3/0416 , G06F3/0488 , G06F2203/04104 , G06F2203/04808 , G06F3/0412 , G06F17/10
Abstract: PURPOSE: A method for extracting multi touch feature information and a multi touch gesture recognizing method using the multi touch feature information are provided to record the features of a multi touch by edge information at a touched point, thereby providing various feature information of the multi touch during touch gesture recognition. CONSTITUTION: Location information about touched points is received from a touch panel(10). Touch points located within a fixed radius around the touched points correspond each other one to one. Edge information is generated wherein the edge information is made of location information of two touch points which are connected each other. Touch graph information is generated and extracted by multi touch feature information. If updated location information is inputted, the touch graph information is updated based on the updated location information.
Abstract translation: 目的：提供一种用于提取多点触摸特征信息的方法和使用多点触摸特征信息的多点触摸手势识别方法，以在触摸点记录多点触摸的边缘信息的特征，从而提供多触摸的各种特征信息 在触摸手势识别期间。 构成：从触摸面板（10）接收关于触摸点的位置信息。 位于触摸点周围的固定半径内的触摸点彼此对应。 产生边缘信息，其中边缘信息由彼此连接的两个触摸点的位置信息组成。 触摸图信息通过多点触控功能信息生成和提取。 如果更新的位置信息被输入，则基于更新的位置信息更新触摸图信息。
-
17.发明授权수치적인 역 운동학 기반의 언센티드 칼만 필터를 이용한 인체 자세 추정 방법, 컴퓨터로 읽을 수 있는 매체 및 서버 시스템 有权使用基于数字反向动画的未经授权的卡尔曼滤波器估计人类位置的方法，计算机可读介质和服务器系统
公开(公告)号：KR101034675B1
公开(公告)日：2011-05-16
申请号：KR1020100098453
申请日：2010-10-08
Applicant: 전남대학교산학협력단
CPC classification number: H04N5/247 , G06T7/75 , H04N5/23219
Abstract: PURPOSE: A human body posture estimating method is provided to estimate accurate human body posture by the reflection of Newton step through a numerical inverse kinematic based UKF(Unscented Kalman Filter). CONSTITUTION: A system calculates a prediction rotation value from the joint rotation value of a first image(S3000). The system calculates a measurement location value from the prediction rotation value(S4000). The system calculates a final measurement rotation value through a UKF(Unscented Kalman Filter)(S5000). The system calculates the final measurement location value from a final observation rotation value(S6000). The system estimates a human body posture(S7000, S8000).
Abstract translation: 目的：提供一种人体姿态估计方法，通过牛顿步反映通过基于数字反运动学的UKF（无声卡尔曼滤波器）来估计人体姿态。 构成：系统根据第一图像的联合旋转值计算预测旋转值（S3000）。 系统根据预测旋转值计算测量位置值（S4000）。 系统通过UKF（无限卡尔曼滤波器）计算最终测量旋转值（S5000）。 系统从最终观测旋转值计算最终测量位置值（S6000）。 系统估计人体姿势（S7000，S8000）。
-
公开(公告)号：KR100933957B1
公开(公告)日：2009-12-28
申请号：KR1020080045490
申请日：2008-05-16
Applicant: 전남대학교산학협력단
IPC: H04N13/02
Abstract: 본 발명은 단일 카메라를 이용한 삼차원 인체 포즈 인식 방법에 관한 것으로, 더욱 상세하게는 서로 다른 포즈를 갖는 기준이 되는 인체들의 실루엣 영상의 특징점에 기반하여 다수 개의 기준 포즈 모델을 생성하여 저장하고, 단일 카메라를 이용하여 검출할 인체의 검출 포즈 모델을 생성하여 상기 기준 포즈 모델들과 비교함으로써 검출할 인체의 삼차원 포즈 및 촬영방향을 추정하는 단일 카메라를 이용한 삼차원 인체 포즈 인식방법에 관한 것이다.
삼차원 포즈, 단일 카메라, 실루엣, 특징점, 히스토그램, 경계영상Abstract translation: 目的：提供一种使用单个相机的三维人体姿态识别方法，通过比较从二维姿态图像获得的角度直方图来估计人体的拍摄方向。 构成：用单个照相机识别的三维人体姿势方法如下。 拍摄具有任意三维姿势的目标物体以获得标准二维姿态图像（S1000）。 提取每个标准二维姿态图像的轮廓图像，并且基于标准轮廓图像创建标准姿态模型。 标准姿态模型存储在标准姿态模型数据库中。 标准姿势模型在重复第一或第三步之后创建，并存储在标准姿势模型数据库中。 获得通过单个相机在任意方向拍摄检测对象的照片的检测二维姿态图像。 基于检测到的轮廓图像和检测对象的视频特征提取检测到的二维姿态图像检测轮廓图像。 获得与检测姿态模型最相似的标准姿态模型。
-
公开(公告)号：KR100853267B1
公开(公告)日：2008-08-20
申请号：KR1020070011105
申请日：2007-02-02
Applicant: 전남대학교산학협력단
Abstract: 본 발명은 스테레오 시각 정보를 이용한 복수 인물 추적 방법 및 그 시스템에 관한 것으로 더욱 상세하게는 스테레오 카메라를 이용하여 인물들 간의 겹침 현상을 효율적으로 구분하고 부분영역 단위로 적분 히스토그램을 이용하여 연산비용을 최소화한 스테레오 시각 정보를 이용한 복수 인물 추적 방법 및 그 시스템에 관한 것이다.
본 발명에 따른 스테레오 시각 정보를 이용한 복수 인물 추적 방법은 스테레오 카메라를 이용하여 배경의 영상을 일정한 초당 프레임으로 촬영하는 단계, 상기 프레임에 움직이는 인물의 영상이 촬영되어 상기 인물의 색상 정보가 추출되는 색상 정보 추출 단계, 상기 추출된 인물의 색상 정보를 이용하여 상기 배경과 상기 인물이 분리되는 단계, 상기 촬영된 인물의 영상을 이용하여 영상 시차를 검출하는 단계, 상기 분리된 인물을 일정한 개수의 부분영역들로 분할하는 단계, 상기 부분영역들 별로 적분 히스토그램을 연산하는 단계, 상기 부분영역들의 히스토그램을 이용하여 객체 모델을 도출하는 단계, 상기 도출된 객체에 이름을 부여하는 레이블링 단계 및 상기 레이블링된 객체를 추적하는 객체 추적 단계를 포함한다.
복수 인물 추적, IHLS, Disparity, 적분 히스토그램, Segmentation-
公开(公告)号：KR101326644B1
公开(公告)日：2013-11-08
申请号：KR1020120087225
申请日：2012-08-09
Applicant: 전남대학교산학협력단
IPC: G06T7/20
CPC classification number: G06T7/277 , G06F17/30 , G06K9/6215 , G06T2210/56
Abstract: The present invention relates to an image tracking method. Particularly, a full-body joint image tracking method using an evolutionary exemplar-based particle filter improves the accuracy of a full-body joint image tracking by determining an optimal pose of the present through a particle filter synthetically based on resampled sample particles after the optimal pose is determined at some previous time and exemplar-based particles renewed along the flow of time. [Reference numerals] (AA) Start;(BB) End;(S1000) Generate exemplar-based pose particle DB and sample particle DB;(S2000) Input a silhouette image;(S3000) Calculate similarity between the silhouette image and exemplar-based pose particles;(S4000) Predict present pose particles from exemplar-based pose particles and sample particles by using a particle filter;(S5000) Calculate similarity between the present pose particles and the silhouette image;(S6000) Determine an optimal pose;(S7000) Re-sample the present pose particles and update sample particle DB;(S8000) Update the optimal pose to the exemplar-based pose particle DB;(S9000) Input a next silhouette image?
Abstract translation: 本发明涉及图像追踪方法。 特别地，使用基于进化的基于示例的粒子滤波器的全身关节图像跟踪方法通过基于重新采样的样本颗粒合成地确定当前通过粒子滤波器的最佳姿态来提高全身关节图像跟踪的精度， 姿势是在以前的某个时间确定的，而基于样本的粒子沿时间流动更新。 （AA）开始;（BB）结束;（S1000）生成基于样本的姿势粒子DB和样本粒子DB;（S2000）输入剪影图像;（S3000）计算轮廓图像和基于样本的相似度 （S4000）通过使用粒子滤波器预测来自基于样本的姿态粒子和样本粒子的姿势粒子（S5000）计算当前姿态粒子与轮廓图像之间的相似度;（S6000）确定最佳姿势;（S7000 ）对目前的姿势粒子进行重新采样并更新采样粒子DB;（S8000）将最佳姿态更新为基于样本的姿态粒子DB;（S9000）输入下一个剪影图像？
-
-
-
-
-
-
-
-
-
Search Results
24
records
(took 0.029 seconds)
