公开(公告)号：US08548802B2

公开(公告)日：2013-10-01

申请号：US12784119

申请日：2010-05-20

申请人： Kazuhiro Nakadai ,  Ince Gokhan ,  Hiroshi Tsujino ,  Tobias Rodemann

发明人： Kazuhiro Nakadai ,  Ince Gokhan ,  Hiroshi Tsujino ,  Tobias Rodemann

IPC分类号： G10L11/04

CPC分类号： G10L21/0208 ,  G10L15/20

摘要： An acoustic data processor according to the present invention is used for processing acoustic data including signal sounds to reduce noises generated by a mechanical apparatus. The acoustic data processor includes a motion status obtaining section for obtaining motion status of the mechanical apparatus, an acoustic data obtaining section for obtaining acoustic data corresponding to the obtained motion status, and a database for storing various motion statuses of the mechanical apparatus in a unit time and corresponding acoustic data as templates. The acoustic data processor further includes a database searching section for searching the database to retrieve the template having the motion status closest to the obtained motion status; and a template subtraction section for subtracting the acoustic data of the template having the motion status closest to the obtained motion status from the obtained acoustic data to reduce noises generated by the mechanical apparatus.

摘要翻译： 根据本发明的声学数据处理器用于处理包括信号声音的声学数据，以减少由机械设备产生的噪声。 声学数据处理器包括用于获得机械装置的运动状态的运动状态获取部分，用于获得与获得的运动状态对应的声学数据的声学数据获取部分，以及用于以单位存储机械装置的各种运动状态的数据库 时间和相应的声学数据作为模板。 声学数据处理器还包括数据库搜索部分，用于搜索数据库以检索具有最接近获得的运动状态的运动状态的模板; 以及模板减法部分，用于从所获得的声学数据中减去具有最接近获得的运动状态的运动状态的模板的声学数据，以减少由机械装置产生的噪声。

公开(公告)号：US20100054489A1

公开(公告)日：2010-03-04

申请号：US12548871

申请日：2009-08-27

申请人： Hirofumi Nakajima ,  Kazuhiro Nakadai ,  Yuji Hasegawa ,  Hiroshi Tsujino

发明人： Hirofumi Nakajima ,  Kazuhiro Nakadai ,  Yuji Hasegawa ,  Hiroshi Tsujino

IPC分类号： H04B3/20

CPC分类号： H04S1/002

摘要： Provided is a dereverberation system or the like which copes with an arbitrary condition flexibly and is capable of recognizing a sound or a sound source signal. According to the dereverberation system, an inverse filter (h) is set by using a pseudo-inverse matrix (R+) of a non-square matrix (R) as a correlation matrix of input signals (x). On the basis of the inverse filter (h) and an estimated correlation matrix (R̂) generated according to a window function (w), an error cost (J(h) between a correlation value of the input signals (x) and output signals (y) and a desired correlation value (d) is calculated. On the basis of the error cost (J(h)), the inverse filter (h) is adaptively updated according to a gradient method.

摘要翻译： 提供了一种能够灵活地处理任意条件并且能够识别声音或声源信号的混响系统等。 根据混响系统，通过使用非正方形矩阵（R）的伪逆矩阵（R +）作为输入信号（x）的相关矩阵来设置逆滤波器（h）。 根据逆滤波器（h）和根据窗函数（w）产生的估计相关矩阵（R），输入信号（x）的相关值和输出信号（x）之间的误差成本（J（h） （y）和期望的相关值（d），基于误差成本（J（h）），逆滤波器（h）根据梯度法自适应地更新。

公开(公告)号：US20080071540A1

公开(公告)日：2008-03-20

申请号：US11900725

申请日：2007-09-12

申请人： Mikio Nakano ,  Kazuhiro Nakadai ,  Hiroshi Tsujino

发明人： Mikio Nakano ,  Kazuhiro Nakadai ,  Hiroshi Tsujino

IPC分类号： G10L15/00

CPC分类号： G10L15/24 ,  G10L15/20

摘要： A robot that recognizes speech of a person while performing predetermined motions or gestures, the robot includes: a drive unit executing the motions or gestures; a determination unit determining one of the motions or gestures being executed; a speech recognition unit having at least two recognition algorithms including a multi-condition training algorithm; and a switch unit selecting one of the recognition algorithms depending on one of the motions or gestures determined.

摘要翻译： 一种机器人，其在执行预定的运动或手势的同时识别人的语音，所述机器人包括：执行所述运动或手势的驱动单元; 确定单元，确定正在执行的运动或手势之一; 语音识别单元，具有包括多条件训练算法的至少两个识别算法; 以及开关单元，其根据确定的运动或手势之一来选择识别算法中的一个。

公开(公告)号：US07295684B2

公开(公告)日：2007-11-13

申请号：US10624786

申请日：2003-07-21

申请人： Hiroshi Tsujino ,  Hiroshi Kondo ,  Shinichi Nagai ,  Koji Akatsuka ,  Atsushi Miura

发明人： Hiroshi Tsujino ,  Hiroshi Kondo ,  Shinichi Nagai ,  Koji Akatsuka ,  Atsushi Miura

IPC分类号： G06K9/00

CPC分类号： G06T7/0002 ,  G06T7/0008 ,  G06T7/215 ,  G06T7/254 ,  G06T7/269

摘要： An object detection apparatus and method capable of detecting objects based on visual images captured by a self-moving unit. A sequential images output section makes a train of a first input image and a second input image sequential to the first input image and outputs said train. A local area image processor calculates local flows based on said first input image and said second input image. An inertia information acquiring section measures self-motion of the unit to calculate inertia information thereof. A global area image processor uses said inertia information to estimate global flow, which is a motion field of the entire view associated to the self-motion, using said global flow and said first input image and creates a predictive image of said second input image. The global area image processor then calculates differential image data, which is a difference between said predictive image and said second input image. A figure-ground segregation section uses said differential image data to refine said local flows and compares the refined local flows with a predetermined threshold value to extract a figure candidate area, which is the area having a high probability of an object existing in the input image. An object presence/absence determination section determines presence/absence of objects in said figure candidate area.

摘要翻译： 一种能够基于由自动移动单元捕获的视觉图像来检测物体的物体检测装置和方法。 顺序图像输出部分构成与第一输入图像顺序的第一输入图像和第二输入图像的列，并输出所述列。 局部区域图像处理器基于所述第一输入图像和所述第二输入图像来计算局部流。 惯性信息获取部分测量该单元的自动运动以计算其惯性信息。 全局区域图像处理器使用所述惯性信息来估计使用所述全局流和所述第一输入图像并创建所述第二输入图像的预测图像的全自动运动的全局流动，其是与自动运动相关联的整个视图的运动场。 然后，全局区域图像处理器计算作为所述预测图像和所述第二输入图像之间的差的差分图像数据。 图形分离部分使用所述差分图像数据来细化所述本地流并将精细局部流与预定阈值进行比较，以提取图形候选区域，其是存在于输入图像中的对象的概率高的区域 。 对象存在/不存在确定部分确定所述图形候选区域中的对象的存在/不存在。

公开(公告)号：US07221797B2

公开(公告)日：2007-05-22

申请号：US10133114

申请日：2002-04-26

申请人： Takamasa Koshizen ,  Koji Akatsuka ,  Hiroshi Tsujino

发明人： Takamasa Koshizen ,  Koji Akatsuka ,  Hiroshi Tsujino

IPC分类号： G06K9/46

CPC分类号： G06K9/00335 ,  G05D1/0253

摘要： An image recognizing apparatus and method is provided for recognizing behavior of a mobile unit accurately with an image of external environment acquired during the mobile unit is moving.Behavior command output block 12 outputs behavior commands to cause the mobile unit 32 move. Local feature extraction block 16 extracts features of local areas of the image from the image of external environment acquired on the mobile unit 32 when the behavior command is output. Global feature extraction block 18 extracts feature of global area of the image using the features of local areas. Learning block 20 calculates probability models for recognizing behavior given to the mobile unit 32 based on the feature of global area of the image. After learning is finished, behavior of the mobile unit 32 may be recognized rapidly and accurately by applying the probability models to an image of external environment acquired in mobile unit 32 afresh.

摘要翻译： 提供一种图像识别装置和方法，用于在移动单元移动期间获取的外部环境的图像精确地识别移动单元的行为。 行为命令输出块12输出行为命令以使移动单元32移动。 局部特征提取块16当输出行为命令时，从移动单元32上获取的外部环境的图像中提取图像的局部区域的特征。 全局特征提取块18使用局部区域的特征提取图像的全局区域的特征。 学习块20基于图像的全局区域的特征来计算用于识别给予移动单元32的行为的概率模型。 在完成学习之后，通过将概率模型应用于移动单元32中获取的外部环境的图像，可以快速且准确地识别移动单元32的行为。

公开(公告)号：US20060280341A1

公开(公告)日：2006-12-14

申请号：US10561256

申请日：2004-06-30

申请人： Takamasa Koshizen ,  Bernd Heisele ,  Hiroshi Tsujino

发明人： Takamasa Koshizen ,  Bernd Heisele ,  Hiroshi Tsujino

IPC分类号： G06K9/00

CPC分类号： G06K9/6255 ,  G06K9/00221 ,  G06K9/00281 ,  G06K9/00973

摘要： A production device and method which produce a multiple-system film having metal components such as TiAlN greatly different in melting point by a melting-evaporation type ion plating method that provides a high material utilization efficiency and a good film quality. Power needed to evaporate a material (4) is first supplied, and then power gradually increased over the initail power is repeatedly supplied until a needed maximum power is reached. Concurrently, a plasma control is performed for converging plasma (7) onto an initial area needed to evaporate the material, and then a plasma control is performed for continuously and sequentially moving/expanding plasma from the initial plasma area up to a maximum plasma area to thereby gradually melt the non-melted portion of the material.

公开(公告)号：US20060184465A1

公开(公告)日：2006-08-17

申请号：US11255895

申请日：2005-10-21

申请人： Hiroshi Tsujino ,  Nobuaki Miyakawa ,  Gen Matsumoto

发明人： Hiroshi Tsujino ,  Nobuaki Miyakawa ,  Gen Matsumoto

IPC分类号： G06N3/02

CPC分类号： G06N3/08 ,  G06N3/049

摘要： A neural network element, outputting an output signal in response to a plurality of input signals, comprises a history memory for accumulating and storing the plurality of input signals in a temporal order as history values. It also includes an output module for outputting the output signal when an internal state exceeds a predetermined threshold value, the internal state being based on a sum of the product of a plurality of input signals and corresponding coupling coefficients. The history values depend on change of the internal state. The neural network element is configured to subtract a predetermined value from the internal state immediately after the output module fires and performs learning for reinforcing or attenuating the coupling coefficient according to the history values after the output module fires.

摘要翻译： 输出响应于多个输入信号的输出信号的神经网络元件包括用于以时间顺序累积并存储多个输入信号作为历史值的历史存储器。 它还包括一个输出模块，用于当内部状态超过预定阈值时输出输出信号，内部状态基于多个输入信号的乘积和相应的耦合系数之和。 历史价值取决于内部状态的变化。 神经网络元件被配置为在输出模块触发之后立即从内部状态减去预定值，并且在输出模块触发之后根据历史值执行用于加强或衰减耦合系数的学习。

公开(公告)号：US20060155660A1

公开(公告)日：2006-07-13

申请号：US10468316

申请日：2002-02-04

申请人： Takamasa Koshizen ,  Hiroshi Tsujino

发明人： Takamasa Koshizen ,  Hiroshi Tsujino

IPC分类号： G06E1/00

CPC分类号： G05B13/0265 ,  G05B13/027

摘要： An agent learning apparatus comprises a sensor (301) for acquiring a sense input, an action controller (307) for creating an action output in response to the sense input and giving the action output to a controlled object, an action state evaluator (303) for evaluating the behavior of the controlled object, a selective attention mechanism (304) for storing the action output and the sense input corresponding to the action output in one of the columns according to the evaluation, calculating a probability model from the action outputs stored in the columns, and outputting, as a learning result, the action output related to a newly given sense input in the column where the highest confidence obtained by applying the newly given sense input to the probability model is stored. By thus learning, the selective attention mechanism (304) obtains a probability relationship between the sense input and the column. An action output is calculated on the basis of the column evaluated as a stable column. As a result, the dispersion of the action output is quickly minimized, and thereby the controlled object can be stabilized.

摘要翻译： 代理学习装置包括用于获取感测输入的传感器（301），用于响应于感测输入创建动作输出并将动作输出给受控对象的动作控制器（307），动作状态评估器（303） 用于评估受控对象的行为的选择性注意机制（304），用于根据评估将动作输出和对应于动作输出的感测输入存储在一列中的选择性注意机制（304），从存储在 并且作为学习结果输出与通过将新给定的感测输入应用于概率模型获得的最高置信度的列中的新给定的感测输入相关的动作输出被存储。 通过这样学习，选择性注意机制（304）获得感知输入和列之间的概率关系。 基于作为稳定列计算的列计算动作输出。 结果，动作输出的分散被快速地最小化，从而可以使受控对象稳定。

公开(公告)号：US07076433B2

公开(公告)日：2006-07-11

申请号：US10052309

申请日：2002-01-17

申请人： Masashi Ito ,  Hiroshi Tsujino

发明人： Masashi Ito ,  Hiroshi Tsujino

IPC分类号： G10L19/02

CPC分类号： G10L21/0272 ,  G10L19/10 ,  G10L21/0264 ,  G10L2021/02166

摘要： A sound separation apparatus for separating a target signal from a mixed input signal, wherein the mixed input signal includes the target signal and one or more sound signals emitted from different sound sources. The sound separation apparatus according comprises a frequency analyzer for performing a frequency analysis on the mixed input signal and calculating spectrum and frequency component candidate points at each time. The apparatus further comprises feature extraction means for extracting feature parameters which are estimated to correspond with the target signal, comprising a local layer for analyzing local feature parameters using the spectrum and the frequency component candidate points and one or more global layers for analyzing global feature parameters using the feature parameters extracted by the local layer. The apparatus further comprises a signal regenerator for regenerating a waveform of the target signal using the feature parameters extracted by the feature extraction means.Since both of local feature parameters and global feature parameters can be processed together in the feature extraction means, the separation accuracy of the target signal is improved without depending on the accuracy for extracting feature parameters from the input signal. Feature parameters to be extracted include frequencies and amplitudes and their variation rates for the frequency component candidate points, harmonic structure, pitch consistency, intonation, on-set/off-set information and/or sound source direction. The number of the layers provided in the feature extraction means may be changed according to the types of the feature parameters to be extracted.

摘要翻译： 一种用于从混合输入信号中分离目标信号的声音分离装置，其中混合输入信号包括目标信号和从不同声源发射的一个或多个声音信号。 声音分离装置包括频率分析器，用于对混合输入信号进行频率分析，并且每次计算频谱和频率分量候选点。 该装置还包括用于提取被估计为与目标信号相对应的特征参数的特征提取装置，包括用于使用频谱和频率分量候选点分析局部特征参数的局部层和用于分析全局特征参数的一个或多个全局层 使用本地层提取的特征参数。 该装置还包括信号再生器，用于使用由特征提取装置提取的特征参数再生目标信号的波形。 由于局部特征参数和全局特征参数可以在特征提取装置中一起处理，所以目标信号的分离精度提高，而不依赖于从输入信号提取特征参数的精度。 要提取的特征参数包括频率分量候选点的频率和幅度及其变化率，谐波结构，音调一致性，语调，定位/偏移信息和/或声源方向。 可以根据要提取的特征参数的类型来改变在特征提取装置中提供的层的数量。

公开(公告)号：US06965694B2

公开(公告)日：2005-11-15

申请号：US09996307

申请日：2001-11-27

申请人： Yugo Ueda ,  Hiroshi Tsujino

发明人： Yugo Ueda ,  Hiroshi Tsujino

IPC分类号： G06K9/00 ,  G06T7/20 ,  G06K9/48

CPC分类号： G06T7/20 ,  G06K9/00335

摘要： A motion information recognition system takes time-series images of an object in motion and generates image vectors that are analyzed to produce a plurality of eigenvectors that are then stored. Inner product operations are performed between image vectors of a recognized object and a plurality of the stored eigenvectors for obtaining the motion information of the recognized object based on the result of said inner product operations to provide high-speed and easy motion information recognition at reduced computing cost.

摘要翻译： 运动信息识别系统拍摄运动中的物体的时间序列图像，并产生被分析以产生多个特征向量然后被存储的图像矢量。 在识别对象的图像向量和多个所存储的本征向量之间执行内部乘积运算，用于基于所述内部乘积运算的结果获得所识别的对象的运动信息，以在减少的计算中提供高速和容易的运动信息识别 成本。