-
公开(公告)号：US5633988A
公开(公告)日：1997-05-27
申请号：US989503
申请日：1992-12-10
CPC分类号： G06N3/08
摘要： An adaptation method for a data processing system comprising a plurality of neurons each of which outputs an output according to a comparison between a sum of multiplied inputs by weights and a threshold, characterized in the threshold of the neuron which has generated significant output at certain point of time is compulsorily increased to a maximal value and the weight of the neuron is adapted for a constant value of the inputs wherein the threshold is decreased to a value at the point.
摘要翻译： 一种用于数据处理系统的适应方法，包括多个神经元，每个神经元根据通过权重的乘法输入和阈值之间的比较来输出输出，其特征在于在特定点产生有效输出的神经元的阈值 的时间被强制地增加到最大值，并且神经元的权重适用于输入的恒定值，其中阈值减小到该点处的值。
-
2.发明授权Method and system for comparing two images by making an initial rough judgement 失效通过初步粗略判断来比较两个图像的方法和系统
公开(公告)号：US5490225A
公开(公告)日：1996-02-06
申请号：US240434
申请日：1994-05-10
申请人： Ryohei Kumagai
发明人： Ryohei Kumagai
摘要： A method and system for comparing an input image with a stored image. An input image is obtained and processed. By the processing of the input image, a rough judgment of whether the input image corresponds to the stored image is obtained. If the input image does correspond to the stored image, then further processing is preformed on the input image. Characteristics of the processed input image are determined. The system then determines whether the characteristics correspond to characteristics of the stored image.
摘要翻译： 一种用于将输入图像与存储的图像进行比较的方法和系统。 获取并处理输入图像。 通过输入图像的处理，可以得到输入图像是否对应于存储图像的粗略判断。 如果输入图像对应于存储的图像，则在输入图像上执行进一步的处理。 确定经处理的输入图像的特性。 然后，系统确定特征是否对应于存储的图像的特性。
-
公开(公告)号：US5355438A
公开(公告)日：1994-10-11
申请号：US51818
申请日：1993-04-26
CPC分类号： G06N3/063
摘要： An analog circuit which performs weighting and thresholding for a neural network. Each neuron of the neural network includes an operational amplifier receiving an input signal, the output of which is connected to a transistor. The transistor conducts only when the output exceeds a predetermined value thereby providing a threshold function. The output of the transistor is connected by a variable resistance to other inputs of other neurons. The output of each operational amplifier thereby corresponds to a weighted version of the input signal, which is adjusted for threshold and is also dependent on other neurons of the network.
摘要翻译： 对神经网络进行加权和阈值处理的模拟电路。 神经网络的每个神经元包括接收输入信号的运算放大器，其输出端连接到晶体管。 晶体管仅在输出超过预定值时导通，从而提供阈值函数。 晶体管的输出通过其他神经元的其他输入的可变电阻连接。 因此，每个运算放大器的输出对应于输入信号的加权版本，其被调整为阈值，并且还取决于网络的其他神经元。
-
公开(公告)号：US5185816A
公开(公告)日：1993-02-09
申请号：US627208
申请日：1990-12-13
CPC分类号： G06K9/6228 , G06N3/02
摘要： A method of selecting characteristics data for a data processing system from a group of input data for reducing data volume of each input data by said data processing system having a neural network structure or a structure equivalent thereto, where each input data consists of a plurality of said characteristics data. The method includes the steps of storing outputs of said input data; selecting a specific characteristics data of a pair of different input data; exchanging said characteristics data of said input data pair with each other; comparing outputs from said data processing system in response to said input data before and after the exchange of said characteristics data; and removing said characteristics data from said input data in said group, when a difference between said outputs before and after is comparatively small.
-
5.发明授权Method and apparatus for identifying a pair of groups in an image having a minimum separation distance 失效用于识别具有最小间隔距离的图像中的一对组的方法和装置
公开(公告)号：US5146549A
公开(公告)日：1992-09-08
申请号：US711493
申请日：1991-06-04
申请人： Ryohei Kumagai
发明人： Ryohei Kumagai
IPC分类号： G06T7/60
CPC分类号： G06T7/606
摘要： A method and apparatus for determining which pair of a plurality of objects in an image has a minimum separation distance. In the method, centroids of groups are ordered in a sequence according to X and Y coordinates. A distance is measured between a reference object and an object adjacent in the ordered sequence to the reference. Objects in the sequence which are not adjacent in the sequence to the reference are excluded from measurement based upon relationships between the coordinate values of the reference object, the adjacent object and the non adjacent object being evaluated for exclusion. This exclusion reduces the total number of steps necessary to determine which pair of objects has the minimum distance. Distance measurements are made to non-excluded objects. Similar steps are repeated for other reference objects. The pair objects which have the smallest measured distance are indicated as the objects having the minimum separation. In the apparatus, the image is stored in a frame memory and an index indicating the centroid of the group may also be stored. The memory may be read in a raster scan order, thus generating an ordered sequence of centroid locations.
摘要翻译： 一种用于确定图像中的多个对象的对具有最小间隔距离的方法和装置。 在该方法中，按照X和Y坐标的顺序排列组的质心。 在参考对象和与有序序列相邻的对象之间测量距离作为参考。 根据参考对象的坐标值，相邻对象和被排除的不相邻对象之间的关系，排除与序列中与参照不相邻的序列中的对象。 这种排除减少了确定哪对物体具有最小距离所需的总步数。 对未排除的物体进行距离测量。 对其他参考对象重复类似的步骤。 具有最小测量距离的对对象被指示为具有最小间隔的对象。 在该装置中，图像被存储在帧存储器中，并且还可以存储指示该组的质心的索引。 可以以光栅扫描顺序读取存储器，从而生成质心位置的有序序列。
-
公开(公告)号：US5136653A
公开(公告)日：1992-08-04
申请号：US501875
申请日：1990-03-29
申请人： Ryohei Kumagai , Sunao Takatori , Makoto Yamamoto
发明人： Ryohei Kumagai , Sunao Takatori , Makoto Yamamoto
摘要： A recognitiion method and system for associating a pattern, such as a printed character with an acoustic signal, such as speech. A multilevel power sequence is generated from the acoustic signal, and the sequence is segmented into phonemes. The multilevel power sequence is converted into a binary level sequence to reduce noise. The binary power sequence is accumulated to give a series whose elements are partial sums of the sequence as a function of time. The accumulated sequence is differentiated to give a series whose elements are the derivatives of the accumulated sequence as a function of time. The differentiated sequence is then associated with a pattern, such as a printed character.
摘要翻译： 一种识别方法和系统，用于将诸如打印字符的图案与声音信号（例如语音）相关联。 从声信号产生多级功率序列，并将该序列分割为音素。 多级功率序列被转换为二进制电平序列以减少噪声。 累积二进制功率序列以给出其序列作为时间的函数的序列的部分和的序列。 累积序列被区分，以给出一个序列，其元素是累积序列的导数作为时间的函数。 差异序列然后与图案（例如打印字符）相关联。
-
公开(公告)号：US5077808A
公开(公告)日：1991-12-31
申请号：US347035
申请日：1989-05-04
申请人： Ryohei Kumagai
发明人： Ryohei Kumagai
CPC分类号： H04N1/4074
摘要： An image processing method for flatten=ing histogram with distribution comprising steps of:a density conversion table is generated for indicating to which densities each density is to be converted, as well as how many pixels are to be converted;the latter of the density is registered in an address designated by the former of the density in a converting circuit, and the number of pixels to be converted is registered in a count-down circuit;density of each pixel of an image is inputted to the converting circuit and the count-down circuit so that latter of the density is outputted from the converting circuit and that the number in an address designated by the density of each pixel is decreased by "1" on each input of the density;the density of each pixel is converted to the former density until the number is decreased to "0"; andpixels in the image of the density which corresponds to the number decreased to "0" is converted to an index density thereafter.
-
公开(公告)号：US5621823A
公开(公告)日：1997-04-15
申请号：US003861
申请日：1993-01-08
申请人： Ryohei Kumagai
发明人： Ryohei Kumagai
摘要： An image processing method for thinning lines converts pixels having a density of a configuration to a background density. Pixels are converted according to conditions which include densities of pixels in a neighborhood around a pixel to be processed. Pixels to be converted to the background density are first converted to an index value. Subsequent pixels are evaluated using the index values of adjacent, previously evaluated pixels. Index values are then converted to the background density.
摘要翻译： 用于减薄线的图像处理方法将具有配置密度的像素转换为背景密度。 根据包括要处理的像素周围的邻域中的像素密度的条件来转换像素。 要转换为背景浓度的像素首先转换为索引值。 使用相邻的先前评估的像素的索引值来评估后续像素。 然后将索引值转换为背景密度。
-
9.发明授权Method for processing data using a neural network having a number of layers equal to an abstraction degree of the pattern to be processed 失效使用具有等于要处理的图案的抽象度的层数的神经网络来处理数据的方法
公开(公告)号：US5553196A
公开(公告)日：1996-09-03
申请号：US461538
申请日：1995-06-05
摘要： There is provided a layer construction of neural layers according to the abstraction degree of data to be processed, and data is inputted to a neural layer corresponding to its abstraction degree.
摘要翻译： 根据待处理数据的抽象度提供神经层的层次构造，并将数据输入到与其抽象度相对应的神经层。
-
公开(公告)号：US5414666A
公开(公告)日：1995-05-09
申请号：US921110
申请日：1992-07-30
申请人： Ryohei Kumagai , Weikang Yang
发明人： Ryohei Kumagai , Weikang Yang
IPC分类号： G06F12/02 , G06F13/16 , G11C11/401 , G11C11/406 , G11C7/00
CPC分类号： G06F13/1636
摘要： A memory control device adaptable to various demands and using a standard DRAM. A memory interface for outputting an address of the memory and controllably reading and writing is connected to the memory. A plurality of input and output ports are connected to the memory interface through a local bus. A host interface is connected to the memory interface through the local bus. A refresh control refreshes the memory through the memory interface. An arbitration structure arbitrates the required access to memory between the refresh control means, the input and output ports and the host interface.
摘要翻译： 适用于各种需求并使用标准DRAM的存储器控制装置。 用于输出存储器的地址和可控读取和写入的存储器接口连接到存储器。 多个输入和输出端口通过本地总线连接到存储器接口。 主机接口通过本地总线连接到存储器接口。 刷新控制通过内存接口刷新内存。 仲裁结构对刷新控制装置，输入和输出端口以及主机接口之间对存储器的所需访问进行了仲裁。
-
-
-
-
-
-
-
-
-
搜索结果
52 条记录 (耗时 0.015 秒)