公开(公告)号：EP0453939A3

公开(公告)日：1993-05-19

申请号：EP91106128.1

申请日：1991-04-17

申请人： YOZAN INC. ,  SHARP KABUSHIKI KAISHA

发明人： Takatori, Sunao ,  Yamamoto, Makoto

IPC分类号： G06F15/80

CPC分类号： G06N3/082

摘要： A learning method for data processing system comprising steps of: i) deciding ignition patterns in input layer and in output layer artificially concerning to plural inputs and outputs; ii) increasing weights of synapses in middle layer and output layer in a case that neurons ignite with more than predetermined rate among those corresponding to neurons in input layer and output layer and neurons adjacent to them in the middle layer; iii) executing the same processing for all inputs and outputs as what is performed above. It is performed in a learning method for data processing system comprising steps of: i) an input layer possessed plural neurons to ignite and output a predetermined data according to result performed a predetermined processing in inputted data; ii) an output layer possessed less neurons than in input layer; and iii) a middle layer, settled between the input layer and output layer, possessed middle number of neurons between those of the input layer and the output layer, and simultaneously, each neuron is combined with each neuron in the input layer and output layer.

公开(公告)号：EP0470530A2

公开(公告)日：1992-02-12

申请号：EP91113070.6

申请日：1991-08-02

申请人： YOZAN INC. ,  SHARP KABUSHIKI KAISHA

发明人： Yamamoto, Makoto

IPC分类号： G07C9/00 ,  G06K9/00 ,  A61B5/117

CPC分类号： G06K9/00067 ,  G07C9/00087

摘要： The invention relates to a fingerprint verification method comprises the steps of:

performing density partial differentials onto a corresponding fingerprint image with respect to 2 directions;
determining a point as a center of characteristics where an extreme value of a histogram of corresponding partial differential value occurs in both directions;
performing a fingerprint verification by only adapting data at the portion upper than the corresponding center of characteristics.

摘要翻译： 本发明涉及一种指纹验证方法，包括以下步骤：相对于两个方向对相应的指纹图像执行密度偏微分; 将点确定为在两个方向上出现相应偏微分值的直方图的极值的特征中心; 通过仅适应在相应的特征中心上方的数据来进行指纹验证。

公开(公告)号：EP0466161A2

公开(公告)日：1992-01-15

申请号：EP91111583.0

申请日：1991-07-11

申请人： YOZAN INC. ,  SHARP KABUSHIKI KAISHA

发明人： Yamamoto, Makoto

IPC分类号： G06K9/46 ,  A61B5/117 ,  G06K9/00

CPC分类号： A61B5/1172 ,  G06K9/00067 ,  G07C9/00087

摘要： An image positioning method, for cocentering two images including ridges each other, comprising the following steps from i) to iv).
i) Calculating density value of each pixel of an image to be examined and a reference image; ii) calculating change of density value between pixels in neighborhood along the first and the second directions of each pixel of the image to be examined and the reference image; iii) calculating histogram of density change along the first and second directions of the image to be examined and the reference image; iv) defining a maximum or minimum point of histogram of the first and second directions to be characteristic point of the image to be examined and the reference image, the image to be examined and the reference image are to be fit so as to the characteristics point coincide each other.

摘要翻译： 一种图像定位方法，用于将包括脊彼此的两个图像对中，包括从i）到iv）的以下步骤。 i）计算要检查的图像的每个像素的密度值和参考图像; ii）计算沿着要检查的图像的每个像素的第一和第二方向的附近的像素之间的浓度值的变化和参考图像; iii）计算沿着待检查图像的第一和第二方向的密度变化的直方图和参考图像; iv）将要被检查的图像的特征点的第一和第二方向的最大或最小直方图定义为参考图像，要被检查的图像和参考图像将适合于特征点 相互重合

公开(公告)号：EP0453939A2

公开(公告)日：1991-10-30

申请号：EP91106128.1

申请日：1991-04-17

申请人： YOZAN INC. ,  SHARP KABUSHIKI KAISHA

发明人： Takatori, Sunao ,  Yamamoto, Makoto

IPC分类号： G06F15/80

CPC分类号： G06N3/082

摘要： A learning method for data processing system comprising steps of: i) deciding ignition patterns in input layer and in output layer artificially concerning to plural inputs and outputs; ii) increasing weights of synapses in middle layer and output layer in a case that neurons ignite with more than predetermined rate among those corresponding to neurons in input layer and output layer and neurons adjacent to them in the middle layer; iii) executing the same processing for all inputs and outputs as what is performed above. It is performed in a learning method for data processing system comprising steps of: i) an input layer possessed plural neurons to ignite and output a predetermined data according to result performed a predetermined processing in inputted data; ii) an output layer possessed less neurons than in input layer; and iii) a middle layer, settled between the input layer and output layer, possessed middle number of neurons between those of the input layer and the output layer, and simultaneously, each neuron is combined with each neuron in the input layer and output layer.

摘要翻译： 一种用于数据处理系统的学习方法，包括以下步骤：i）在人为地对多个输入和输出确定输入层和输出层中的点火模式; ii）在输入层和输出层与中间层相邻的神经元相对应的神经元对应神经元中，神经元以超过预定速率点燃的情况下，增加中层和输出层突触的重量; iii）对所有输入和输出执行与上述相同的处理。 在数据处理系统的学习方法中进行包括以下步骤：i）具有多个神经元的输入层，根据在输入数据中执行预定处理的结果点火并输出预定数据; ii）输出层具有比输入层少的神经元; 和iii）在输入层和输出层之间设置的中间层，在输入层和输出层之间具有中间数量的神经元，同时，每个神经元与输入层和输出层中的每个神经元组合。

公开(公告)号：EP0450522A2

公开(公告)日：1991-10-09

申请号：EP91105034.2

申请日：1991-03-28

申请人： YOZAN INC. ,  SHARP KABUSHIKI KAISHA

发明人： Takatori, Sunao ,  Yamamoto, Makoto

IPC分类号： G06F15/80

CPC分类号： G06K9/4628 ,  G06K2209/01 ,  G06N3/082 ,  Y10S706/90

摘要： A learning method for data processing system characterized in: i) deciding ignition patterns in input layer and in output layer artificially concerning to plural inputs and outputs; ii) increasing weights of synapses in middle layer and output layer in a case that neurons ignite with more than certain rate among those corresponding to neurons in input layer and output layer and neurons adjacent to them in the middle layer; iii) executing the same processing for all inputs and outputs as what is performed above; in a learning method for data processing system comprising: i) an input layer and an output layer possessed plural neurons to ignite and output a certain data according to result performed a certain processing in inputted data; and ii) a middle layer, settled between the input layer and output layer, possessed the plural neurons and simultaneously, each neuron is combined with each neuron in the input layer and output layer.

摘要翻译： 一种用于数据处理系统的学习方法，其特征在于：i）人为地确定与多个输入和输出有关的输入层和输出层中的点火模式; ii）在输入层和输出层中的神经元以及中间层中与其相邻的神经元对应的神经元中，神经元以一定比率点燃的情况下，增加中层和输出层中的突触的权重; iii）对于所有输入和输出执行与上面执行的相同的处理; 在数据处理系统的学习方法中，包括：i）输入层和输出层具有多个神经元以根据输入数据中执行的特定处理的结果来点燃并输出特定数据; 和ii）位于输入层和输出层之间的中间层具有多个神经元，并且每个神经元与输入层和输出层中的每个神经元同时结合。

公开(公告)号：EP0420254A3

公开(公告)日：1991-07-10

申请号：EP90118592.6

申请日：1990-09-27

申请人： YOZAN INC. ,  SHARP KABUSHIKI KAISHA

发明人： Takatori, Sunao ,  Kumagai, Ryohei ,  Matsumoto, Koji ,  Yamamoto, Makoto

IPC分类号： G06F15/80

CPC分类号： G06N3/04

摘要： A data processing system comprising a plurality of neural layers characterized in that a part of neurons in one of the layers are connected to neurons in the following layer.

公开(公告)号：EP0420254A2

公开(公告)日：1991-04-03

申请号：EP90118592.6

申请日：1990-09-27

申请人： YOZAN INC. ,  SHARP KABUSHIKI KAISHA

发明人： Takatori, Sunao ,  Kumagai, Ryohei ,  Matsumoto, Koji ,  Yamamoto, Makoto

IPC分类号： G06F15/80

CPC分类号： G06N3/04

摘要： A data processing system comprising a plurality of neural layers characterized in that a part of neurons in one of the layers are connected to neurons in the following layer.

摘要翻译： 一种包括多个神经层的数据处理系统，其特征在于，其中一层中的一部分神经元连接到下一层中的神经元。