公开(公告)号：US20070244842A1

公开(公告)日：2007-10-18

申请号：US11579981

申请日：2005-06-02

申请人： Mie Ishii ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Yusuke Mitarai

发明人： Mie Ishii ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Yusuke Mitarai

IPC分类号： G06F15/18

CPC分类号： G06K9/4628 ,  G06N3/04 ,  G06N3/082 ,  G06T7/12 ,  G06T2207/20084 ,  G06T2207/30201

摘要： An output value of neuron within an objective layer of a hierarchical neural network is computed. The data of the output value of neuron is stored in a memory only if the output value of neuron is greater than or equal to a predetermined value by referring to the computed output value of neuron within the objective layer. When the data of the output value of neuron on a former layer of objective layer is read from the memory, the data having a predetermined value is read, instead of the data of the output value of neuron not stored in the memory.

摘要翻译： 计算层次神经网络目标层内神经元的输出值。 如果神经元的输出值大于或等于预定值，则通过参照计算出的目标层内神经元的输出值，将神经元的输出值的数据存储在存储器中。 当从存储器中读取目标层之前的神经元的输出值的数据时，读取具有预定值的数据，而不是未存储在存储器中的神经元的输出值的数据。

公开(公告)号：US20070162828A1

公开(公告)日：2007-07-12

申请号：US10580327

申请日：2005-01-24

申请人： Yusuke Mitarai ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Takashi Morie

发明人： Yusuke Mitarai ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Takashi Morie

IPC分类号： H03M13/00

CPC分类号： H04N19/192 ,  H04N19/119 ,  H04N19/136 ,  H04N19/172 ,  H04N19/50 ,  H04N19/60

摘要： Data in multidimensional space such as a two-dimensional image is encoded with high efficiency. Further, as two-dimensional data can be decomposed to one-dimensional bases, the problem of wiring for two-dimensional parallelizing in a convolution arithmetic unit can be solved. For this purpose, two-dimensional image data f(x,y) to be encoded is inputted, and one-dimensional adaptive bases X(x),Y(y) representing the two-dimensional image are obtained. Next, a reconstructed image is generated based on the one-dimensional adaptive bases, and the one-dimensional adaptive bases are corrected based on an error E between the reconstructed image and the input image. The correction is repeated until the error E is reduced.

摘要翻译： 多维空间中的数据（如二维图像）以高效率进行编码。 此外，由于二维数据可以分解为一维的基础，因此可以解决卷积运算单元中二维并行化的布线问题。 为此，输入要编码的二维图像数据f（x，y），并且获得表示二维图像的一维自适应基准X（x），Y（y）。 接下来，基于一维自适应基础生成重建图像，并且基于重构图像和输入图像之间的误差E来校正一维自适应基准。 重复校正直到误差E减小。

公开(公告)号：US08755611B2

公开(公告)日：2014-06-17

申请号：US13207143

申请日：2011-08-10

申请人： Shunta Tate ,  Yusuke Mitarai ,  Hiroto Yoshii ,  Yasuhiro Okuno ,  Katsuhiko Mori ,  Masakazu Matsugu

发明人： Shunta Tate ,  Yusuke Mitarai ,  Hiroto Yoshii ,  Yasuhiro Okuno ,  Katsuhiko Mori ,  Masakazu Matsugu

IPC分类号： G06K9/68

CPC分类号： G06K9/6282

摘要： An information processing apparatus that discriminates the orientation of a target includes a calculation unit that calculates a distribution of a difference in feature amount between a plurality of learning patterns each showing the orientation of a target, a determination unit that determines, using a probability distribution obtained from the distribution of differences calculated by the calculation unit, a pixel that is in an input pattern and is to be referred to in order to discriminate the orientation of a target in the input pattern, and a discrimination unit that performs discrimination for obtaining the orientation of the target in the input pattern by comparing a feature amount of the pixel determined by the determination unit and a threshold set in advance.

摘要翻译： 鉴别目标的方向的信息处理装置包括计算单元，该计算单元计算显示目标的取向的多个学习模式之间的特征量的差异的分布;确定单元，其使用获得的概率分布来确定 根据由计算单元计算出的差异的分布，输入图案中要被标识为输入图案中的目标的方位的像素，以及进行用于获得取向的辨别的判别单元 通过比较由确定单元确定的像素的特征量和预先设定的阈值来输入图案中的目标。

公开(公告)号：US08015130B2

公开(公告)日：2011-09-06

申请号：US12696993

申请日：2010-01-29

申请人： Masakazu Matsugu ,  Katsuhiko Mori ,  Mie Ishii ,  Yusuke Mitarai

发明人： Masakazu Matsugu ,  Katsuhiko Mori ,  Mie Ishii ,  Yusuke Mitarai

IPC分类号： G06E1/00 ,  G06E3/00 ,  G06F15/18 ,  G06G7/00

CPC分类号： G06K9/4628 ,  G06K9/00604 ,  G06N3/049 ,  G06N3/08

摘要： In a hierarchical neural network having a module structure, learning necessary for detection of a new feature class is executed by a processing module which has not finished learning yet and includes a plurality of neurons which should learn an unlearned feature class and have an undetermined receptor field structure by presenting a predetermined pattern to a data input layer. Thus, a feature class necessary for subject recognition can be learned automatically and efficiently.

摘要翻译： 在具有模块结构的分层神经网络中，由尚未完成学习的处理模块执行用于检测新特征类所需的学习，并且包括多个神经元，其应该学习未知的特征类并具有未确定的受体场 通过向数据输入层呈现预定图案来构造。 因此，可以自动且有效地学习用于对象识别所需的特征类。

公开(公告)号：US07991719B2

公开(公告)日：2011-08-02

申请号：US11579981

申请日：2005-06-02

申请人： Mie Ishii ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Yusuke Mitarai

发明人： Mie Ishii ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Yusuke Mitarai

IPC分类号： G06N3/00

CPC分类号： G06K9/4628 ,  G06N3/04 ,  G06N3/082 ,  G06T7/12 ,  G06T2207/20084 ,  G06T2207/30201

摘要： An output value of neuron within an objective layer of a hierarchical neural network is computed. The data of the output value of neuron is stored in a memory only if the output value of neuron is greater than or equal to a predetermined value by referring to the computed output value of neuron within the objective layer. When the data of the output value of neuron on a former layer of objective layer is read from the memory, the data having a predetermined value is read, instead of the data of the output value of neuron not stored in the memory.

摘要翻译： 计算层次神经网络目标层内神经元的输出值。 如果神经元的输出值大于或等于预定值，则通过参照计算出的目标层内神经元的输出值，将神经元的输出值的数据存储在存储器中。 当从存储器中读取目标层之前的神经元的输出值的数据时，读取具有预定值的数据，而不是未存储在存储器中的神经元的输出值的数据。

公开(公告)号：US20110150286A1

公开(公告)日：2011-06-23

申请号：US12956037

申请日：2010-11-30

申请人： Miho Ishigami ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Yusuke Mitarai

发明人： Miho Ishigami ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Yusuke Mitarai

IPC分类号： G06K9/00

CPC分类号： G06K9/00993 ,  G06T7/77 ,  G06T2207/10028

摘要： An estimation apparatus for estimating a position and orientation of an object, includes: a capturing unit adapted to capture an object targeted for position and orientation estimation, and generate a range image representing distance information from the capturing unit to the target object; a general estimation unit adapted to analyze the range image and estimate a general position and orientation of the target object; a plurality of identification units each adapted to estimate a detailed position and orientation of an object within a predetermined position and orientation range; a determination unit adapted to determine a priority order of the plurality of identification units, based on the general position and orientation estimated by the general estimation unit; and a detailed estimation unit adapted to estimate a detailed position and orientation of the target object, using the plurality of identification units in the priority order determined by the determination unit.

摘要翻译： 一种用于估计对象的位置和姿态的估计装置，包括：拍摄单元，其适于拍摄针对位置和姿态估计的对象，并且生成表示从拍摄单元到目标对象的距离信息的距离图像; 适用于分析所述范围图像并估计所述目标对象的一般位置和方位的通用估计单元; 多个识别单元，其各自适于估计在预定位置和取向范围内的物体的详细位置和取向; 确定单元，其适于基于由所述一般估计单元估计的一般位置和方向来确定所述多个识别单元的优先级顺序; 以及详细估计单元，其适于使用由所述确定单元确定的优先级顺序来使用所述多个识别单元来估计所述目标对象的详细位置和方位。

公开(公告)号：US07480584B2

公开(公告)日：2009-01-20

申请号：US10580327

申请日：2005-01-24

申请人： Yusuke Mitarai ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Takashi Morie

发明人： Yusuke Mitarai ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Takashi Morie

IPC分类号： G01D1/00 ,  G06F15/00 ,  G06M11/04

CPC分类号： H04N19/192 ,  H04N19/119 ,  H04N19/136 ,  H04N19/172 ,  H04N19/50 ,  H04N19/60

摘要： Data in multidimensional space such as a two-dimensional image is encoded with high efficiency. Further, as two-dimensional data can be decomposed to one-dimensional bases, the problem of wiring for two-dimensional parallelizing in a convolution arithmetic unit can be solved. For this purpose, two-dimensional image data f(x,y) to be encoded is inputted, and one-dimensional adaptive bases X(x),Y(y) representing the two-dimensional image are obtained. Next, a reconstructed image is generated based on the one-dimensional adaptive bases, and the one-dimensional adaptive bases are corrected based on an error E between the reconstructed image and the input image. The correction is repeated until the error E is reduced.

摘要翻译： 多维空间中的数据（如二维图像）以高效率进行编码。 此外，由于二维数据可以分解为一维的基础，因此可以解决卷积运算单元中二维并行化的布线问题。 为此，输入要编码的二维图像数据f（x，y），并且获得表示二维图像的一维自适应基准X（x），Y（y）。 接下来，基于一维自适应基础生成重建图像，并且基于重构图像和输入图像之间的误差E来校正一维自适应基准。 重复校正直到误差E减小。

公开(公告)号：US07088860B2

公开(公告)日：2006-08-08

申请号：US10105309

申请日：2002-03-26

申请人： Masakazu Matsugu ,  Katsuhiko Mori ,  Osamu Nomura

发明人： Masakazu Matsugu ,  Katsuhiko Mori ,  Osamu Nomura

IPC分类号： G06K9/00

CPC分类号： G06K9/4628 ,  G06K9/00281 ,  G06K9/00973 ,  G06N3/049 ,  G06N3/063

摘要： A plurality of signal processing functions are achieved with the same arithmetic processing circuit by controlling wiring arrangements or signal modulation in accordance with a predetermined arrangement control signal that is output based on circuit arrangement information read from a circuit arrangement information storage unit. Hierarchical parallel processing is realized with small-scale circuit configuration. Further, detection of a predetermined feature and integration of the detection results can be efficiently performed.

公开(公告)号：US20060074653A1

公开(公告)日：2006-04-06

申请号：US10530275

申请日：2004-12-16

申请人： Yusuke Mitari ,  Masakazu Matsugu ,  Katsuhiko Mori

发明人： Yusuke Mitari ,  Masakazu Matsugu ,  Katsuhiko Mori

IPC分类号： G10L15/08

CPC分类号： G06K9/00248 ,  G06K9/4628

摘要： Pattern recognition capable of robust identification for the variance of an input pattern is performed with a low processing cost while the possibility of identification errors is decreased. In a pattern recognition apparatus which identifies the pattern of input data from a data input unit (11) by using a hierarchical feature extraction processor (12) which hierarchically extracts features, an extraction result distribution analyzer (13) analyzes a distribution of at least one feature extraction result obtained by a primary feature extraction processor (121). On the basis of the analytical result, a secondary feature extraction processor (122) performs predetermined secondary feature extraction.

摘要翻译： 以低处理成本执行能够对输入模式的方差进行鲁棒识别的模式识别，同时识别错误的可能性降低。 在通过使用分层提取特征的分层特征提取处理器（12）来识别来自数据输入单元（11）的输入数据的模式的模式识别装置中，提取结果分布分析器（13）分析至少一个 特征提取结果由主要特征提取处理器（121）获得。 基于分析结果，次要特征提取处理器（122）执行预定的二次特征提取。

公开(公告)号：US20050280711A1

公开(公告)日：2005-12-22

申请号：US11145380

申请日：2005-06-03

申请人： Mie Ishii ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Yusuke Mitarai

发明人： Mie Ishii ,  Masakazu Matsugu ,  Katsuhiko Mori ,  Yusuke Mitarai

IPC分类号： H04N5/225 ,  G06F17/00 ,  H04N5/232

CPC分类号： H04N5/232

摘要： Camera information corresponding to the position at which the target to be searched for in the captured image is detected is identified, another camera specified by the identified camera information receives a notification that the target has been detected (S209), and when a notification is received from another camera, the posture of the own camera is controlled so that the area associated with the camera information specifying another camera can be observed (S211).

摘要翻译： 识别对应于检测到拍摄图像中要搜索的目标的位置的相机信息，由所识别的相机信息指定的另一相机接收到已经检测到目标的通知（S 209），并且当通知为 从另一相机接收，控制自身相机的姿态，使得可以观察到与指定另一相机的相机信息相关联的区域（S 211）。