专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明专利

JP2008262288A Estimation device 有权
标题翻译：估计装置
公开(公告)号：JP2008262288A
公开(公告)日：2008-10-30
申请号：JP2007102915
申请日：2007-04-10
申请人： Denso Corp , Nara Institute Of Science & Technology , 国立大学法人奈良先端科学技術大学院大学 , 株式会社デンソー
发明人： SHIMIZU MIKIRO , FUKAYA NAOKI , BANDO YOSHIJI , SHIBATA TOMOHIRO , ISHII MAKOTO
IPC分类号： G06T7/60 , G01B11/26 , G06T1/00 , G06T7/00 , G06T7/20
摘要： PROBLEM TO BE SOLVED: To provide an estimation device capable of obtaining stable estimation results just after the start of the processing and enhancing an estimation accuracy.
SOLUTION: A face model providing portion 6 provides an average face model S
A stored in an average face model storage portion 61 to an estimation portion 8 estimating an affine parameter A
t for obtaining a head pose via a face model switching portion 63 when an individual face model learning portion 62 starts in response to an initialization instruction. The individual face model learning portion 62 acquires a result of tracking characteristic points (observed values) on the image by the estimation portion 8, learns an individual face model S
P , terminates learning the individual face model when a free energy of the individual face model S
P is over a free energy of the average face model S
A , and switches a face model C
a provided to the estimation portion 8 from the average face model S
A to the individual face model S
P .
COPYRIGHT: (C)2009,JPO&INPIT
摘要翻译：要解决的问题：提供一种能够在开始处理之后获得稳定的估计结果并提高估计精度的估计装置。解决方案：面部模型提供部分6将存储在平均面部模型存储部分61中的平均面部模型S A 提供给估计部分8，估计仿射参数A SB>，用于当个人面部模型学习部分62响应初始化指令开始时通过面部模型切换部分63获得头部姿势。个人面部模型学习部分62通过估计部分8获取跟踪特征点（观察值）的结果，学习个人面部模型S P ，终止学习个人面部模型时个人脸部模型S P 的自由能超过平均脸部模型S A 的自由能，并切换面部模型C a 从平均面部模型S A 提供给估计部分8到个人面部模型S P 。版权所有（C）2009，JPO＆INPIT

2. 发明专利

JP2008261662A Three-dimensional shape restoration device 有权
标题翻译：三维形状恢复装置
公开(公告)号：JP2008261662A
公开(公告)日：2008-10-30
申请号：JP2007102914
申请日：2007-04-10
申请人： Denso Corp , Nara Institute Of Science & Technology , 国立大学法人奈良先端科学技術大学院大学 , 株式会社デンソー
发明人： SHIMIZU MIKIRO , FUKAYA NAOKI , BANDO YOSHIJI , SHIBATA TOMOHIRO , ISHII MAKOTO
IPC分类号： G01B11/24 , G06T1/00
摘要： PROBLEM TO BE SOLVED: To execute factorization even if low-reliability data exist in an observation matrix (a feature-point position series), as to a three-dimensional shape restoration device.
SOLUTION: In factorizing the observation matrix W, a reliability matrix Q is used which expresses the reliability (variance of probability distribution (normal distribution) expressing the position of a feature point) of respective observation values x
t constituting the observation matrix W. By causing the variance Σ
s of test distribution q(S) of a shape matrix to reflect the reliability matrix Q, factorization is performed in which the higher the reliability of feature points is, the more they are emphasized.
COPYRIGHT: (C)2009,JPO&INPIT
摘要翻译：要解决的问题：即使在观察矩阵（特征点位置序列）中存在低可靠性数据，也要执行因子分解，关于三维形状恢复装置。解决方案：在对观测矩阵W进行因子分解时，使用可靠性矩阵Q，其表示各个观测值x 的可靠性（表示特征点的位置的概率分布的方差（正态分布） SB>构成观察矩阵W.通过使形状矩阵的测试分布q（S）的方差ΣS S / S S反映可靠性矩阵Q，进行因子分解，其中特征的可靠性越高分数越多，他们就越强调。版权所有（C）2009，JPO＆INPIT

3. 发明专利

JP2011053798A Driving environment determination device, driving situation estimation device and driving support device 有权
标题翻译：驾驶环境决定装置，驾驶状况估计装置及驾驶支援装置
公开(公告)号：JP2011053798A
公开(公告)日：2011-03-17
申请号：JP2009200369
申请日：2009-08-31
申请人： Denso Corp , Nara Institute Of Science & Technology , 国立大学法人奈良先端科学技術大学院大学 , 株式会社デンソー
发明人： FUKAYA NAOKI , BANDO YOSHIJI , HITOMI KENTARO , IKEDA KAZUSHI , SHIBATA TOMOHIRO , AJIMA HIROYOSHI
IPC分类号： G08G1/16
摘要： PROBLEM TO BE SOLVED: To more properly support driving according to a driver. SOLUTION: A driving environment determination part 11 selects driving environment determination cluster distribution based on distribution selection information (identification information ID, weather information W, time zone information TB), and when defining specified manipulated variables Hsb based on a vehicle speed V, steering manipulated variables S and brake manipulated variables B as a parameter for determination, estimates a driving environment by collating the parameter for determination with the selected environment determination cluster distribution. A target value setting part 15 sets target values OD and OT of an inter-vehicle distance and the time to collision suitable for the current driving environment or a driver, weather, and time zone by using an estimation model prepared for each driving environment. A driving situation estimation part 16 outputs the degree of deviation of actual detection values D and TTC corresponding to the target values OD and OT as an estimation value showing the driving situations of the driver. COPYRIGHT: (C)2011,JPO&INPIT
摘要翻译：要解决的问题：更适当地支持驾驶者驾驶。解决方案：驾驶环境确定部分11基于分配选择信息（识别信息ID，天气信息W，时区信息TB）来选择驾驶环境确定集群分布，并且当基于车速V定义指定的操纵变量Hsb时转向操纵变量S和制动操作变量B作为确定参数，通过将确定参数与所选择的环境确定集群分布进行比较来估计驾驶环境。目标值设定部15通过使用为每个行驶环境准备的估计模型，设定车辆间距离的目标值OD和OT以及适合当前驾驶环境或驾驶员，天气和时区的碰撞时间。驱动情况估计部分16输出与目标值OD和OT相对应的实际检测值D和TTC的偏差程度作为表示驾驶员驾驶状况的估计值。版权所有（C）2011，JPO＆INPIT

4. 发明专利

JP2013178827A Driving situation estimation device and driving support system 有权
标题翻译：驾驶状况估计装置和驾驶支持系统
公开(公告)号：JP2013178827A
公开(公告)日：2013-09-09
申请号：JP2013117260
申请日：2013-06-03
申请人： Denso Corp , 株式会社デンソー , Nara Institute Of Science & Technology , 国立大学法人奈良先端科学技術大学院大学
发明人： FUKAYA NAOKI , BANDO YOSHIJI , HITOMI KENTARO , IKEDA KAZUSHI , SHIBATA TOMOHIRO , AJIMA HIROYOSHI
IPC分类号： G08G1/16 , B60R21/00
摘要： PROBLEM TO BE SOLVED: To more accurately perform driving support in accordance with a driver.SOLUTION: In a driving environment determination section 11, an environment determination cluster distribution is selected according to distribution selection information (identification information ID, weather information W, time band information TB), a standardized manipulated variable Hsb based on a car velocity V, a steering manipulated variable S and a brake manipulated variable B is used as a determination parameter, and the determination parameter is collated with the selected environment determination cluster distribution, thereby estimating a driving environment. In a target value setting section 15, an estimation model prepared for each driving environment is used to set target values OD, OT of a vehicular gap and a collision tolerance time suitable for a current driving environment, driver, weather and time band. A driving situation estimation section 16 outputs a degree of deviation of actual detection values D, TTC with respect to the target values OD, OT as an estimate indicating the driving situation of the driver.
摘要翻译：要解决的问题：根据驾驶员更准确地执行驾驶辅助。解决方案：在驾驶环境确定部分11中，根据分配选择信息（识别信息ID，天气信息W，时间带）选择环境确定集群分布信息TB），基于汽车速度V的标准化操作变量Hsb，转向操纵变量S和制动操作变量B作为确定参数，并且将所述确定参数与所选择的环境确定集群分布进行整理，从而估计一个驾驶环境。在目标值设定部15中，使用为每个驾驶环境准备的估计模型来设定车辆间隙的目标值OD，OT，适合于当前行驶环境，驾驶员，天气和时间带的碰撞容许时间。驾驶状态估计部分16输出相对于目标值OD，OT的实际检测值D，TTC的偏差程度，作为指示驾驶员的行驶状况的估计。

5. 发明专利

JP2010123070A Three-dimensional shape restoration device 有权
标题翻译：三维形状恢复装置
公开(公告)号：JP2010123070A
公开(公告)日：2010-06-03
申请号：JP2008298478
申请日：2008-11-21
申请人： Denso Corp , Nara Institute Of Science & Technology , 国立大学法人奈良先端科学技術大学院大学 , 株式会社デンソー
发明人： HITOMI KENTARO , BANDO YOSHIJI , FUKAYA NAOKI , SHIBATA TOMOHIRO , IKEDA KAZUSHI
IPC分类号： G06T1/00 , G01B11/245
摘要： PROBLEM TO BE SOLVED: To execute factorization even if data originating from a plurality of objects exist in an observation data matrix (a feature-point position series), in a three-dimensional shape restoration device. SOLUTION: In factorizing the observation data matrix x showing the positions of the feature points originating from the plurality of objects, a shape matrix showing the shapes of the objects is assumed as z, a specifying matrix for specifying an object to which each feature point belongs is assumed as h, a motion matrix of an object u is assumed as W e¾u , measurement accuracy of the feature point originating from the object u is assumed as τ ¾u , a model parameter obtained by summarizing probability (mixing ratio) g in which the feature point belongs to each object u (u=1, 2, ..., U) is assumed as θ, and a variation Bayes estimation method is applied. A second test distribution q(z, h) is determined while a first test distribution q(θ) is fixed in an E step, and the first test distribution q(θ) is determined while the second test distribution q(z, h) is fixed in an M step, thereby probabilistically executing factorization calculation. COPYRIGHT: (C)2010,JPO&INPIT
摘要翻译：要解决的问题：即使在三维形状恢复装置中，即使在观察数据矩阵（特征点位置系列）中存在来自多个对象的数据，也执行因式分解。解决方案：在对表示从多个对象产生的特征点的位置的观察数据矩阵x进行分解时，将表示对象的形状的形状矩阵假设为z，用于指定每个对象的对象的指定矩阵将特征点属于假设为h，将对象u的运动矩阵假定为W e¾u，则将来自对象u的特征点的测量精度设为τ¾u

6. 发明专利

JP2007172237A Estimation device 有权
标题翻译：估计装置
公开(公告)号：JP2007172237A
公开(公告)日：2007-07-05
申请号：JP2005368124
申请日：2005-12-21
申请人： Denso Corp , Nara Institute Of Science & Technology , 国立大学法人奈良先端科学技術大学院大学 , 株式会社デンソー
发明人： FUKAYA NAOKI , SHIMIZU MIKIRO , ISHII MAKOTO , SHIBATA TOMOHIRO , BANDO YOSHIJI
IPC分类号： G06T7/20 , G06T1/00
CPC分类号： G06K9/00248
摘要： PROBLEM TO BE SOLVED: To provide an estimation device, for accurately performing time-series Bayes estimation in real time even in an environment with non-Gaussian noise.
SOLUTION: A mixed distribution generation part 14 generates a particle according to a mixed distribution in which an upper prediction distribution based on an estimation result of an affine parameter showing a head posture and a lower prediction distribution based on an estimation result of a face characteristic point are mixed in a mixing ratio α
a, t , α
z, t , and an observation part 22 calculates the weight of the particle using observation data z
t and a template tp
n , whereby an estimated distribution of the position of the face characteristic point expressed by weighted particles is determined. Based on particles sampled along the estimation distribution and both the prediction distributions used for generation of the mixed distribution, a mixing ratio calculation part 20 calculates a mixing ratio determined later so as to provide a mixing distribution most approximate to the estimated distribution as a mixing ratio α
a, t+1 , α
z, t+1 used in the next estimation.
COPYRIGHT: (C)2007,JPO&INPIT
摘要翻译：要解决的问题：为了提供一种估计装置，即使在具有非高斯噪声的环境中也能实时准确地执行时间序列贝叶斯估计。解决方案：混合分布生成部分14根据混合分布生成粒子，其中基于表示头部姿势的仿射参数的估计结果的上部预测分布和基于a的估计结果的较低预测分布表面特征点以混合比α a，t ，αSB> t 混合，观察部22使用观测数据z t 和模板tp n ，由此确定由加权粒子表示的面部特征点的位置的估计分布。基于沿着估计分布采样的颗粒和用于产生混合分布的预测分布，混合比计算部20计算稍后确定的混合比，以便提供最接近估计分布的混合分布作为混合比 α a，t + 1 ，α z，t + 1 。版权所有（C）2007，JPO＆INPIT

7. 发明专利

JP2013037471A Probabilistic model update system, probabilistic model update device, probabilistic model update method, and program 审中-公开
标题翻译：概率模型更新系统，概念模型更新装置，概念模型更新方法和程序
公开(公告)号：JP2013037471A
公开(公告)日：2013-02-21
申请号：JP2011171836
申请日：2011-08-05
申请人： Nec Corp , 日本電気株式会社 , Nara Institute Of Science & Technology , 国立大学法人奈良先端科学技術大学院大学
发明人： KAMIYA YUKI , KUNIEDA KAZUO , SHIBATA TOMOHIRO , HAYASHI KOHEI , TAKENOUCHI TAKASHI , IKEDA KAZUSHI
IPC分类号： G06F19/00 , G06F17/18
摘要： PROBLEM TO BE SOLVED: To make measurement information having three or more dimensions a highly accurate probabilistic model and to allow the probabilistic model to be successively updated by stream processing.SOLUTION: A reception part 11 receives sensor information showing users' behaviors from two or more sensors that detect behaviors of two or more users, and a sensor information storage part 12 stores the sensor information. A modelling part 13 provides a probabilistic model of the sensor information stored to the sensor information storage part 12 on the assumption that the user and the sensor are not changed with a time as a variable. A model storage part 14 stores the probabilistic model. A model update part 15 updates the probabilistic model stored to the model storage part 14 on the basis of sensor information that the modelling part 13 does not use for providing the probabilistic model. A missing value prediction part 16 and an unexpected value detection part 17 compare the sensor information with the probabilistic model, predict missing values respectively, and detect unexpected values.
摘要翻译：要解决的问题：使具有三维或更多维度的测量信息是高度精确的概率模型，并允许通过流处理连续地更新概率模型。解决方案：接收部分11从两个或更多个检测两个或更多个用户的行为的传感器接收表示用户行为的传感器信息，传感器信息存储部分12存储传感器信息。假设用户和传感器不以时间作为变量而改变，建模部分13提供存储到传感器信息存储部分12的传感器信息的概率模型。模型存储部分14存储概率模型。模型更新部分15基于建模部分13不用于提供概率模型的传感器信息来更新存储到模型存储部分14的概率模型。缺失值预测部分16和意外值检测部分17将传感器信息与概率模型进行比较，分别预测缺失值并检测意外值。版权所有（C）2013，JPO＆INPIT

8. 发明专利

JP2010237729A Subjective evaluation value detection device, subjective evaluation value detection method and program 有权
标题翻译：主观评估价值检测装置，主观评价价值检测方法及程序
公开(公告)号：JP2010237729A
公开(公告)日：2010-10-21
申请号：JP2009081748
申请日：2009-03-30
申请人： Nara Institute Of Science & Technology , Nec Corp , 国立大学法人奈良先端科学技術大学院大学 , 日本電気株式会社
发明人： KAMIYA YUKI , KATO HIROSHI , KUNIEDA KAZUO , SHIBATA TOMOHIRO , IKEDA KAZUSHI , KITA IZUMI
IPC分类号： G06Q10/00 , G06Q50/00 , G06Q50/10
摘要： PROBLEM TO BE SOLVED: To provide a subjective evaluation value detection device for evaluating the degree of subjective view of a person from information on switch operation based on his or her sense. SOLUTION: The switch 101 in intuitively operated by a user subjectively for a predetermined sense. A feature amount extraction unit 102 extracts input operation information that is a set of predetermined feature amounts from the input operation to the switch 101. An input operation information storage unit 103 accumulates the input operation information. An input operation profile information generation unit 104 determines an evaluation axis where the average information amount (entropy) of the combination of predetermined feature amounts of the user's input operation based on the input operation information. An evaluation value calculation unit 106 calculates an evaluation value by projecting the input operation information to the evaluation axis determined by the input operation profile information generation unit 104. COPYRIGHT: (C)2011,JPO&INPIT
摘要翻译：要解决的问题：提供一种主观评估值检测装置，用于根据他或她的感觉从关于切换操作的信息来评估人的主观观看度。
解决方案：开关101由用户主观地以预定的感觉直观地操作。特征量提取单元102从输入操作向开关101提取作为一组预定特征量的输入操作信息。输入操作信息存储单元103累积输入的操作信息。输入操作简档信息生成单元104基于输入的操作信息来确定用户输入操作的预定特征量的组合的平均信息量（熵）的评估轴。评估值计算单元106通过将输入的操作信息投影到由输入操作简档信息生成单元104确定的评估轴来计算评估值。版权所有（C）2011，JPO＆INPIT

9. 发明专利

JP2009066693A Motion data generating device, method, and robot device 审中-公开
标题翻译：运动数据生成装置，方法和机器人装置
公开(公告)号：JP2009066693A
公开(公告)日：2009-04-02
申请号：JP2007236629
申请日：2007-09-12
申请人： Nara Institute Of Science & Technology , Research Organization Of Information & Systems , Toyota Motor Corp , トヨタ自動車株式会社 , 国立大学法人奈良先端科学技術大学院大学 , 大学共同利用機関法人情報・システム研究機構
发明人： RI KAIKEN , INAMURA TETSUYA , SHIBATA TOMOHIRO
IPC分类号： B25J9/22 , B25J5/00
摘要： PROBLEM TO BE SOLVED: To provide a motion data generating device which generates various motion patterns from a predetermined motion pattern, and also to provide a motion pattern generating method, a program, and a robot device.
SOLUTION: The motion data generating device has: a symbol generating section 12 which generates a new source symbol from two or more source symbols in a source symbol space where similar information between the source symbols is held, upon projection of the source symbols, based on a gap amount between the same; and a pattern generating section 13 which generates an output time sequence pattern from a state transition probability and an output probability of the new source symbol. The source symbol corresponds to a predetermined motion by which the state transition probability and the output probability are calculated by using a hidden Markov model, and the symbol generating section 12 converts the state transition probabilities of the two or more source symbols into information concerning a speed, and calculates the state transition probability of the new source symbol.
COPYRIGHT: (C)2009,JPO&INPIT
摘要翻译：要解决的问题：提供一种从预定运动模式生成各种运动模式的运动数据生成装置，并且还提供运动模式生成方法，程序和机器人装置。运动数据产生装置具有：符号生成部分12，其在源符号空间中从源符号空间中的两个或更多个源符号生成新的源符号，其中源符号的投影被保持在源符号之间，，基于相同的间隙量; 以及模式生成部13，其从新的源符号的状态转移概率和输出概率生成输出时间序列模式。源符号对应于通过使用隐马尔可夫模型来计算状态转移概率和输出概率的预定运动，并且符号生成部12将两个或更多个源符号的状态转移概率转换为关于速度的信息，并计算新的源符号的状态转移概率。版权所有（C）2009，JPO＆INPIT

10. 发明专利

JP2014059654A Dynamic body three-dimensional motion detection device and detection method 有权
标题翻译：动态体积三维运动检测装置及检测方法
公开(公告)号：JP2014059654A
公开(公告)日：2014-04-03
申请号：JP2012203488
申请日：2012-09-14
申请人： Nara Institute Of Schience And Technology , 国立大学法人奈良先端科学技術大学院大学
发明人： SHIBATA TOMOHIRO , FUNATANI HIROYUKI , NAKAMURA TERUHIRO
IPC分类号： G06T7/20 , G01B11/00
摘要： PROBLEM TO BE SOLVED: To provide a device and a method capable of automatically tracking three-dimensional motions of feature points such as limbs and a head part of a small animal such as a mouse, which can apply not only to a small animal but also to a general animal such as a human and a non-biological dynamic body such as a robot.SOLUTION: A dynamic body three-dimensional motion detection device detects each point on a surface of a dynamic body 12 moving on the opposite side via a shield plate 11 which shields visible light but transmits infrared with an infrared depth sensor 13 in chronological order. The device connects adjacent points of each point detected at each time point with connection lines, expresses the surface of the dynamic body 12 with a mesh of connection lines. Feature points of the dynamic body 12 at each time point are extracted from the mesh, and the three-dimensional motion of the dynamic body is detected by the chronological change thereof.
摘要翻译：要解决的问题：提供能够自动跟踪特征点的三维运动的装置和方法，所述特征点例如小鼠的四肢和头部，其不仅可以应用于小动物，而且可以应用于小动物涉及一般动物如人类和非生物动力体如机器人。解决方案：动态身体三维运动检测装置通过屏蔽物检测在相对侧上移动的动态体12的表面上的每个点板11屏蔽可见光，但按时间顺序用红外深度传感器13透射红外。设备将在每个时间点检测到的每个点的相邻点与连接线连接，用连接线的网格表示动态主体12的表面。从网格提取每个时间点的动态体12的特征点，并且通过其时间顺序的变化来检测动态体的三维运动。

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式