-
公开(公告)号：KR101178176B1
公开(公告)日：2012-08-29
申请号：KR1020100137882
申请日：2010-12-29
Applicant: 성균관대학교산학협력단
Abstract: 본 발명은 확률적 다중 해석 생성에 기반한 3D 물체 인식 및 자세 추정 시스템 및 그 방법에 대한 것으로서, 특히 약한 인식기를 통해 생성된 물체의 자세 후보들에 융합된 확률 증거를 추가하여 보다 정확하게 물체를 인식하고 그 자세를 추정할 수 있는 확률적 다중 해석 생성에 기반한 3D 물체 인식 및 자세 추정 시스템 및 그 방법에 관한 것이다. 본 발명은 약한 인식기를 이용하여 획득된 불확실한 증거를 기초로 물체의 자세 후보들을 생성하고, 이에 추가적인 확률 증거를 추가하여 보다 정확하게 물체를 인식하고 자세를 추정할 수 있는 확률적 다중 해석 생성에 기반한 3D 물체 인식 및 자세 추정 시스템 및 그 방법을 제공할 수 있다.
-
公开(公告)号：KR1020120044792A
公开(公告)日：2012-05-08
申请号：KR1020100106288
申请日：2010-10-28
Applicant: 성균관대학교산학협력단
Abstract: PURPOSE: An outlier removal method for recognition based visual tracking is provided to estimate a posture of an object according to comparison between a tracking point prediction result and a real next frame tracking point. CONSTITUTION: If an extracted feature point of a designated object is grater than predetermined number, the feature point is generated as a tracking point(S120,S130). A tracking thread determines the kind of the tracking point(S140). If the number of stored inlier tracking points is grater than a predetermined number, the tracking thread reflects the inlier tracking point on a posit algorithm(S160,S170). The tracking thread estimates a posture of the object.
Abstract translation: 目的：提供基于识别的视觉跟踪的异常值去除方法，以根据跟踪点预测结果和实际下一帧跟踪点之间的比较来估计对象的姿势。 构成：如果指定对象的提取特征点比预定数目更大，则生成特征点作为跟踪点（S120，S130）。 跟踪线确定跟踪点的种类（S140）。 如果存储的反射点跟踪点的数量比预定数量更多，则跟踪线程在一个定位算法上反映该入射点跟踪点（S160，S170）。 跟踪线程估计对象的姿势。
-
公开(公告)号：KR1020120042278A
公开(公告)日：2012-05-03
申请号：KR1020100103899
申请日：2010-10-25
Applicant: 성균관대학교산학협력단
Abstract: PURPOSE: A cognitive emotion expression device of a robot using a fuzzy Petri net is provided to improve the cognitive functions of a robot because the place and transition of the Petri net are set differently depending on the personality of the robot. CONSTITUTION: A cognitive emotion expression device(100) of a robot using a fuzzy Petri net comprises a cognition unit(110), an emotion management unit(120), and an actuator(130). The cognition unit recognizes an external stimulus. The emotion management unit determines a cognitive emotion corresponding to the external stimulus and selects an action depending on the determined cognitive emotion. The actuator is operated depending on the selected action.
Abstract translation: 目的：提供一种使用模糊Petri网的机器人的认知情感表达装置，以提高机器人的认知功能，因为Petri网的位置和转换根据机器人的个性而不同。 构成：使用模糊Petri网的机器人的认知情感表达装置（100）包括认知单元（110），情绪管理单元（120）和致动器（130）。 认知单位认识到外部刺激。 情感管理单元确定与外部刺激相对应的认知情绪，并根据确定的认知情绪选择一个动作。 执行器根据所选择的动作进行操作。
-
公开(公告)号：KR101124878B1
公开(公告)日：2012-03-27
申请号：KR1020100039450
申请日：2010-04-28
Applicant: 성균관대학교산학협력단
Abstract: 종래의 영상의 흐림반경을 추정하는 방법과 비교하여 영상의 흐림반경값을 보다 정확히 예측할 수 있고 영상에 존재하는 노이즈에도 강한 특성을 가지는 추정된 흐림반경에 기초한 영상보정방법 및 영상보정장치가 개시되어 있다. 영상의 흐림반경을 추정하여 영상을 보정하는 영상보정방법에 있어 영상데이타를 제공받는 단계와 제공된 영상의 흐림에 관한 데이터값과 영상의 흐림 정도를 모델링한 오차함수간의 차이를 구하여 제공된 영상의 흐림반경(Blur Radius)를 추정하는 단계를 이용해 영상의 흐림반경을 추정하여 영상을 보정할 수 있다. 따라서, 정확한 영상의 흐림반경을 추정하여 영상촬영장치 또는 디스플레이장치 등에서 영상을 보정할 경우 실제의 영상데이타에 가까운 영상으로 보정할 수 있다.
-
公开(公告)号：KR1020120026864A
公开(公告)日：2012-03-20
申请号：KR1020100089036
申请日：2010-09-10
Applicant: 성균관대학교산학협력단
Inventor： 이석한
Abstract: PURPOSE: A context intersection based line matching method is provided to extract a feature point by using a line intersection context feature even when the texture of image is insufficient. CONSTITUTION: A line is extracted from a consecutive image(S1-1). A intersection point is acquired from an extracted line(S1-2). A set of feature points is defined(S1-3). A LICF(Line Intersection Context Feature) matching is performed(S2-1-1). Mismatches are eliminated by RANSAC(RANdom SAmple Consensus)(S2-1-2). A line component is matched(S2-2).
Abstract translation: 目的：提供基于上下文交点的线匹配方法，即使图像纹理不足，也可以使用线交叉上下文特征来提取特征点。 构成：从连续图像中提取一行（S1-1）。 从提取线获取交点（S1-2）。 定义一组特征点（S1-3）。 执行LICF（线路交叉口上下文特征）匹配（S2-1-1）。 RANSAC（RANdom SAmple Consensus）（S2-1-2）消除了不匹配。 线分量匹配（S2-2）。
-
Patent Agency Ranking
公开(公告)号：KR1020110119989A
公开(公告)日：2011-11-03
申请号：KR1020100039450
申请日：2010-04-28
Applicant: 성균관대학교산학협력단
Abstract: PURPOSE: An image compensation method based on presumed overcast radius and image compensation apparatus thereof are provided to obtain clear image similar with a real image by predicting a function value. CONSTITUTION: An image compensation apparatus predicts a function value in which is modeled the blurred edge of an image(1010). The image compensation apparatus predicts μ values and σ values based on predicted A and B value(1020). The image compensation apparatus deducts the predicted A and B value based on the predicted μ values and σ values(1030). The image compensation apparatus eliminates the overcast image through an estimated overcast radius value(1050).
Abstract translation: 目的：提供一种基于推定的阴影半径的图像补偿方法及其图像补偿装置，以通过预测功能值来获得与实像相似的清晰图像。 构成：图像补偿装置预测其中模拟图像的模糊边缘的功能值（1010）。 图像补偿装置基于预测的A和B值（1020）来预测μ值和σ值。 图像补偿装置根据预测的μ值和σ值（1030）减去预测的A和B值。 图像补偿装置通过估计阴影半径值消除阴影图像（1050）。
-
公开(公告)号：KR101054736B1
公开(公告)日：2011-08-05
申请号：KR1020100041642
申请日：2010-05-04
Applicant: 성균관대학교산학협력단
IPC: H04N13/00
CPC classification number: G06K9/6212 , G06K9/6215 , G06K2209/40 , G06T5/40 , H04N13/204
Abstract: PURPOSE: A 3D object recognition and pose measuring method are provided to measure similarity with a target model with a D image by using a mult-part HSV color histogram. CONSTITUTION: A 2D image is obtained through a stereo camera(110). A 3D point cloud is generated from the 2D image(115). Similarity between a color histogram of image patches in the 2D image and a color histogram of a target model is measured(135). Based on the similarity, a plurality of image patches among the image patches are selected(160). A pose hypothesis corresponding to the selected image patches is predicted(165).
Abstract translation: 目的：提供3D对象识别和姿态测量方法，通过使用多部分HSV颜色直方图来测量与D图像的目标模型的相似度。 构成：通过立体相机（110）获得2D图像。 从2D图像生成3D点云（115）。 测量2D图像中图像斑块的颜色直方图与目标模型的颜色直方图之间的相似性（135）。 基于相似度，选择图像块中的多个图像块（160）。 预测与所选择的图像块相对应的姿势假设（165）。
-
公开(公告)号：KR1020110079541A
公开(公告)日：2011-07-07
申请号：KR1020100137882
申请日：2010-12-29
Applicant: 성균관대학교산학협력단
Abstract: PURPOSE: A three-dimensional object recognition and posture estimation system based on probabilistic multi-analysis and a method thereof are provided to accurately recognize an object and estimate the posture using additional probability evidences. CONSTITUTION: A three-dimensional object recognition and posture estimation method based on probabilistic multi-analysis is as follows. The feature of an object is extracted from a single or plural images inputted from an image acquisition device. The candidates for the posture of the object are created from the extracted object feature. The probability of actual existence of the object is assigned to each posture candidate. The pairs of posture candidates are integrated between the posture candidates which are created based on the features of the object extracted before and after the observation of the image acquisition device. The final posture candidate is selected, and the actual probability of the object is updated.
Abstract translation: 目的：提供一种基于概率多分析的三维物体识别和姿势估计系统及其方法，用于准确识别物体，并使用附加的概率证据来估计姿势。 构成：基于概率多分析的三维物体识别和姿态估计方法如下。 从从图像获取装置输入的单个或多个图像中提取对象的特征。 从提取的对象特征创建对象姿势的候选者。 将对象的实际存在的概率分配给每个姿势候选。 姿势候补对被集成在基于在图像获取装置的观察之前和之后提取的对象的特征而创建的姿势候选之间。 选择最终姿势候选者，并更新对象的实际概率。
-
公开(公告)号：KR100927335B1
公开(公告)日：2009-11-19
申请号：KR1020080017750
申请日：2008-02-27
Applicant: 성균관대학교산학협력단
IPC: G06T17/00
Abstract: A self modeling method of a 3D rotating symmetrical object and an apparatus thereof are provided to grasp the whole shape of an object by estimating the geometric shape of the object from partial data. The main axis of a 3D point cloud is estimated(S10). The 3D point cloud is divided into 3D points belonging to each plane. The center points of each plane are obtained. A center point cloud is obtained(S20). An intrinsic vector having the biggest eigenvalue is selected as a new main axis(S30). When the previous main axis and new main axis exist within a threshold value, the new main axis is selected as the direction vector of the center axis of the 3D object(S42).
Abstract translation: 提供3D旋转对称物体的自建模方法及其装置，以通过从部分数据估计物体的几何形状来掌握物体的整体形状。 估计3D点云的主轴（S10）。 3D点云被分成属于每个平面的3D点。 获得每个平面的中心点。 获得中心点云（S20）。 选择具有最大特征值的固有矢量作为新的主轴（S30）。 当先前主轴和新主轴存在于阈值内时，选择新主轴作为3D目标的中心轴的方向矢量（S42）。
-
公开(公告)号：KR1020090081921A
公开(公告)日：2009-07-29
申请号：KR1020080008100
申请日：2008-01-25
Applicant: 성균관대학교산학협력단
CPC classification number: B41J2/16 , B41J2/1628 , B41J2/1629 , B41J2/1631
Abstract: An ink jetting apparatus using electrostatic force, a manufacturing method thereof, and an ink providing method thereof are provided to facilitate the selection of ink regardless of color change due to heat as a heater is unnecessary by spraying ink with static power generated by electric potential difference between upper electrode and lower electrode. An ink jetting apparatus using electrostatic force comprises a lower electrode part(100), an upper electrode part(200), and a bonding layer(300). In the lower electrode part, a nozzle formed on the top of a first substrate, a lower electrode positioned inside the nozzle and an ink inflow path formed at the bottom of the first substrate. The upper electrode part comprises an upper electrode formed on the top of a second substrate and an ink discharging hole perforated from the bottom of the second substrate to the upper electrode. The bonding layer bonds the lower electrode and the upper electrode to arrange the nozzle and the ink discharging hole in a vertical direction. Meniscus is formed in the nozzle tip when static force generated by the electric potential difference between the upper electrode and the lower electrode is applied on the ink supplied to the nozzle through the ink inflow path and the ink having micro droplet size is discharged through the ink discharging hole of the upper electrode at the end of the meniscus.
Abstract translation: 提供一种使用静电力的喷墨设备及其制造方法和墨水提供方法，以便随着加热器的颜色变化而无论颜色变化如何，通过喷涂由电位差产生的静电而不需要墨水 在上电极和下电极之间。 使用静电力的喷墨装置包括下电极部（100），上电极部（200）和接合层（300）。 在下部电极部中，形成在第一基板的顶部上的喷嘴，位于喷嘴内的下部电极和形成在第一基板的底部的墨水流入路径。 上电极部分包括形成在第二基板的顶部上的上电极和从第二基板的底部到上电极穿孔的排墨孔。 接合层结合下电极和上电极，以使喷嘴和排墨孔在垂直方向上布置。 当由通过墨水流入路径供给到喷嘴的墨水上施加由上部电极和下部电极之间的电位差产生的静电力时，在喷嘴尖端形成弯液面，并且具有微滴尺寸的墨水通过墨水排出 在半月板的末端放出上电极的孔。
-
-
-
-
-
-
-
-
-
Search Results
101
records
(took 0.034 seconds)
