专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

21. 发明专利

JP2011243154A Device and method for detecting vehicle 审中-公开
标题翻译：用于检测车辆的装置和方法
公开(公告)号：JP2011243154A
公开(公告)日：2011-12-01
申请号：JP2010117280
申请日：2010-05-21
申请人： Mitsubishi Electric Corp , 三菱電機株式会社
发明人： MIYAHARA HIROYASU , YONEYAMA SHOGO , HIRANO TAKASHI , NIIYAMA MARIKA , INOUE SATORU
IPC分类号： G08G1/16 , B60R1/00 , G06T1/00 , G06T7/20
摘要： PROBLEM TO BE SOLVED: To provide a vehicle detection device that prevents erroneous detection of a stationary object, such as a guardrail or a sidewall, as another vehicle (an approaching object) approaching the vehicle on which the device is mounted, even when a consistent distance is kept between the vehicle and the stationary object, e.g., a guardrail or sidewall, on a straight road and the relative velocity therebetween is approximately zero.SOLUTION: The vehicle detection device extracts movement information indicating the moving direction and size of an object from input image data that is acquired by imaging the outside of the vehicle on which the device is mounted, and determines whether a stationary object exists based on regularity of the extracted movement information. Therefore, the vehicle detection device can perform control to prevent erroneous detection of a stationary object, such as a sidewall, as another vehicle.
摘要翻译：要解决的问题：提供一种车辆检测装置，其防止诸如护栏或侧壁的静止物体的错误检测，作为接近安装有该装置的车辆的另外的车辆（接近物体）甚至当在车辆和固定物体（例如护栏或侧壁）之间在直路上保持一致的距离并且其间的相对速度大约为零时。车辆检测装置从通过对安装有该装置的车辆的外部成像而获取的输入图像数据提取表示物体的移动方向和尺寸的移动信息，并且基于该静止物体是否存在提取运动信息的规律性。因此，车辆检测装置可以进行控制，以防止作为其他车辆的诸如侧壁的静止物体的错误检测。版权所有（C）2012，JPO＆INPIT

22. 发明专利

JP2011180687A Multilingual document analysis device 审中-公开
标题翻译：多文档分析设备
公开(公告)号：JP2011180687A
公开(公告)日：2011-09-15
申请号：JP2010042321
申请日：2010-02-26
申请人： Mitsubishi Electric Corp , 三菱電機株式会社
发明人： HIRANO TAKASHI , MIKAMI TAKASHI , KAWAMATA TAKENORI
IPC分类号： G06K9/62
摘要： PROBLEM TO BE SOLVED: To provide a multilingual document analysis device capable of quickly and determining the type of the language of a character written in an image in a document with high accuracy.
SOLUTION: The type of language of a text extracted from an electronic document 101 is determined, and the type of language in performing the character recognition of the character written in the image is selected from the determination result, and the character recognition of the image extracted from the electronic document 101 is performed on the basis of the type of language selected, and the type of language of the character written in the image is determined from the character recognition result.
COPYRIGHT: (C)2011,JPO&INPIT
摘要翻译：要解决的问题：提供能够快速地确定以高精度在文档中写入图像中的字符的语言的类型的多语言文档分析装置。解决方案：从电子文档101提取的文本的语言类型被确定，并且从确定结果中选择执行写入图像的字符的字符识别的语言的类型，并且字符识别从电子文档101提取的图像根据所选择的语言的类型进行，并且根据字符识别结果确定写入图像中的字符的语言的类型。版权所有（C）2011，JPO＆INPIT

23. 发明专利

JP2011175568A Inter-document distance calculator and text retriever 有权
标题翻译：文档间距计算器和文本检索
公开(公告)号：JP2011175568A
公开(公告)日：2011-09-08
申请号：JP2010040578
申请日：2010-02-25
申请人： Mitsubishi Electric Corp , 三菱電機株式会社
发明人： MIKAMI TAKASHI , HIRANO TAKASHI
IPC分类号： G06F17/30 , G06F17/21
摘要： PROBLEM TO BE SOLVED: To solve such problems that a correct similarity cannot be obtained, a similarity between synonyms and between texts of different expressions cannot be calculated with accuracy, and further a similar text following the similarity of a text structure cannot be classified when a co-occurrence between unrelated words with each other exists in a text with a plurality of subjects, even though the similarity between documents and between the document and the retrieval key during retrieval is calculated on the basis of a co-occurrence of words in a text.
SOLUTION: A syntax analyzing means performs a morphological analysis and a dependency analysis of a character string of a document from a document inputting means, a tree structure with syntax information creating means creates a tree structure with syntax information from the syntax analysis result, a parallel node adding means adds a parallel node with nodes having a parallel relation to the tree structure with syntax information defined as a child node to the tree structure with syntax information, a parallel node ordering means orders nodes under the added parallel node, the tree structure with syntax information is edited into a tree structure with syntax information of another document, and a distance calculating means calculates the edit distance.
COPYRIGHT: (C)2011,JPO&INPIT
摘要翻译：要解决的问题：为了解决无法获得正确相似性的问题，不能准确地计算不同表达式的同义词和文本之间的相似度，并且进一步类似于文本结构相似的文本不能是在不相关词之间的同现存在于具有多个主题的文本中时，即使文档之间的相似性和在检索期间的文档与检索关键字之间的相似性是基于词的共现来计算的在一个文本。解决方案：语法分析装置从文档输入装置执行文档的字符串的形态分析和依赖性分析，具有语法信息创建装置的树结构从语法分析结果创建具有语法信息的树结构并行节点添加装置将具有与树结构具有平行关系的节点的并行节点与被定义为子节点的语法信息相加用于具有语法信息的树结构，并行节点排序装置对所添加的并行节点下的节点进行排序，将具有语法信息的树结构编辑为具有另一文档的语法信息的树结构，并且距离计算装置计算编辑距离。版权所有（C）2011，JPO＆INPIT

24. 发明专利

JP2009276826A Information retrieval device and information retrieval program 审中-公开
标题翻译：信息检索设备和信息检索程序
公开(公告)号：JP2009276826A
公开(公告)日：2009-11-26
申请号：JP2008124798
申请日：2008-05-12
申请人： Mitsubishi Electric Corp , 三菱電機株式会社
发明人： KOMICHI YUSUKE , HIRANO TAKASHI , WATANABE KEISUKE , TANIGAKI KOICHI
IPC分类号： G06F17/30
摘要： PROBLEM TO BE SOLVED: To provide an information retrieval device for improving adaptability and reproducibility by using general information in the specialized field of a user for retrieval request extension without preliminarily describing it in an ontology.
SOLUTION: An initial section structure generation part 1 generates an initial section structure model from the retrieval request of a user. A relevant section structure generation part 2 acquires the initial section structure model as general information in the specialized field of the user from a meta database 6 for retrieval request extension, and for extending it to the relevant section structure model. A section structure matching meta data retrieval part 3 retrieves meta data whose section structure is matched with that of the relevant section structure model from a retrieval target meta database 7, and a retrieval result output part 4 presents a matched document.
COPYRIGHT: (C)2010,JPO&INPIT
摘要翻译：要解决的问题：提供一种信息检索装置，用于通过使用用于检索请求扩展的用户的专业领域中的一般信息来提高适应性和可重现性，而无需在本体中初步描述。解决方案：初始部分结构生成部分1根据用户的检索请求生成初始部分结构模型。相关部分结构生成部分2从用于检索请求扩展的元数据库6获取初始部分结构模型作为用户的专用字段中的一般信息，并将其扩展到相关的部分结构模型。匹配元数据检索部分3的部分结构从检索目标元数据库7中检索其分段结构与相关部分结构模型的分段结构匹配的元数据，并且检索结果输出部分4呈现匹配的文档。版权所有（C）2010，JPO＆INPIT

25. 发明专利

JP2005235099A Metadata extraction device 有权
标题翻译：元数据提取装置
公开(公告)号：JP2005235099A
公开(公告)日：2005-09-02
申请号：JP2004046611
申请日：2004-02-23
申请人： Mitsubishi Electric Corp , 三菱電機株式会社
发明人： KAMESHIRO TAIZOU , HIRANO TAKASHI
IPC分类号： G06F17/21 , G06F17/28 , G06F17/30
摘要： PROBLEM TO BE SOLVED: To provide a metadata extraction device capable of extracting useful knowledge of relations between sentences/words in a document, or the like. SOLUTION: The metadata extraction device is provided with; a document acquisition means 11 which acquires an electronic document; a character extraction means 12 which extracts characters constituting sentences and relative position data of each character within the electronic document, from the acquired electronic document; an attribute definition DB 17 wherein an attribute of a character string consisting of one or more characters constituting sentences is defined and stored in association with parts of speech of characters constituting the character string; an attribute assignment means 13 which refers to attribute definitions to assign an attribute to the character string; an inter-attribute relevancy calculation means 14 which calculates a relevancy between noticed character strings out of character strings to which attributes are assigned; and an attribute structuring means 15 which gives attributes assigned by the attribute assignment means 13 to respective character strings as identifiers and divides character strings into groups in accordance with relevancies calculated by the inter-attribute relevancy calculation means 14 and gives corresponding identifiers to pertinent groups to generate structured information. COPYRIGHT: (C)2005,JPO&NCIPI
摘要翻译：要解决的问题：提供能够提取文档中的句子/单词之间的关系的有用知识的元数据提取装置等。解决方案：提供元数据提取设备; 获取电子文档的文档获取装置11; 字符提取装置12，从获取的电子文档中提取组成句子的文字和电子文档中每个字符的相对位置数据; 属性定义DB 17，其中由构成句子的一个或多个字符组成的字符串的属性与构成字符串的字符的语音的部分相关联地被定义和存储; 属性分配装置13，其引用属性定义以将属性分配给字符串; 属性相关性计算装置14，其计算从分配了属性的字符串中的注意字符串之间的相关性; 以及属性构造装置15，其将由属性分配装置13分配的属性赋予各自的字符串作为标识符，并且根据由属性间相关性计算装置14计算的相关性将字符串划分成组，并将相应的标识符赋予相关组，生成结构化信息。版权所有（C）2005，JPO＆NCIPI

26. 发明专利

JP2004265452A Character recognition device 有权
公开(公告)号：JP2004265452A
公开(公告)日：2004-09-24
申请号：JP2004188254
申请日：2004-06-25
申请人： Mitsubishi Electric Corp , 三菱電機株式会社
发明人： HIRANO TAKASHI , OKADA YASUHIRO , YODA FUMIO
IPC分类号： G06K9/40 , G06T5/00 , H04N1/409
摘要： PROBLEM TO BE SOLVED: To solve the problem that a table or a character string on a document picture is erroneously broken. SOLUTION: A roller noise removing part 16b is provided which counts the number of black pixels existing in the document picture for each vertical line and compares counted values of respective vertical lines with a threshold from both right and left ends toward the center of the document picture and determines an area ending with the first position where the counted values are smaller than the threshold, as a roller noise area and substitutes black pixels included in the roller noise area with white pixels. Thus the position of roller noise can be accurately obtained, and as a result, the problem that a table or a character string on the document picture is erroneously broken is solved. COPYRIGHT: (C)2004,JPO&NCIPI

27. 发明专利

JP2014153846A Pattern recognizer 有权
标题翻译：图案识别装置
公开(公告)号：JP2014153846A
公开(公告)日：2014-08-25
申请号：JP2013022008
申请日：2013-02-07
申请人： Mitsubishi Electric Corp , 三菱電機株式会社
发明人： HIRANO TAKASHI , ISHII JUN
IPC分类号： G06K9/68
摘要： PROBLEM TO BE SOLVED: To improve recognition accuracy in the dimensional compression of the extracted characteristic vector of an input pattern in pattern recognition and to solve a problem in that LDA of a characteristic dimension compression system limits the number of dimensions of a compression characteristic vector and a dimension with fewer categories and a lower number of dimensions results in less expression in the characteristic vector and unsatisfactory recognition accuracy.SOLUTION: A characteristic extraction part extracts a characteristic vector from the correlation data of a pattern group having: a pattern of categories of a recognition target data of a pattern group of a recognition target, and non-recognition target data that is a non-recognition target although it is correlated with the recognition target and that is similar to the recognition target data in terms of category; the pattern of a category of non-recognition target data erroneously input when pattern recognition takes place; and noise pattern. An in-group/inter-group square sum product sum matrix calculation part obtains an in-group square sum product sum matrix and an inter-group square sum product sum matrix from extracted characteristics of two types. A characteristic dimension compression matrix part calculates a characteristic dimension compression matrix from the in-group square sum product sum matrix and the inter-group square sum product sum matrix.
摘要翻译：要解决的问题：为了提高模式识别中输入图案的提取特征向量的尺寸压缩中的识别精度，并且解决了特征尺寸压缩系统的LDA限制压缩特性矢量的维数的问题，以及具有较少类别和较小维数的维度导致特征向量中的较少表达和不令人满意的识别精度。解决方案：特征提取部分从具有以下各项的模式组的相关数据中提取特征向量：识别目标的图案组的识别目标数据和作为非识别目标的非识别目标数据虽然与识别目标相关，并且与类别中的识别目标数据类似; 在进行模式识别时错误地输入的一类不识别目标数据的模式; 和噪音模式。组内/组间平方和乘积和矩阵计算部分从两种类型的提取特征中获得组内平方和积和矩阵和群间平方和积和矩阵。特征尺寸压缩矩阵部分根据组内平方和乘积和矩阵和组间平方和乘积和矩阵计算特征尺寸压缩矩阵。

28. 发明专利

JP2013097591A Alarm device for obstacle on rear and right/left side of vehicle 审中-公开
标题翻译：用于在后方和右/左侧车辆上发生障碍的报警装置
公开(公告)号：JP2013097591A
公开(公告)日：2013-05-20
申请号：JP2011240156
申请日：2011-11-01
申请人： Mitsubishi Electric Corp , 三菱電機株式会社
发明人： HIRANO TAKASHI , NIIYAMA MARIKA , HASHIMO TETSUJI , KATAYAMA MASAHIKO , TAKEHARA NARIAKI
IPC分类号： G08G1/16 , G06T1/00
摘要： PROBLEM TO BE SOLVED: To solve the problem that a conventional device which monitors the rear and right/left sides of a vehicle, which are dead zones for a driver of the vehicle, to warn the driver on detection of another vehicle has erroneously detected guard rails, wall surfaces, pedestrians on sidewalks, or the like other than object vehicles in some cases and any of erroneous detection suppression methods has some defects.SOLUTION: A white line detection processing unit analyzes an image captured by an in-vehicle camera to detect white lines dividing lanes on the right and the left of a vehicle of a user. A white line geometric position calculation processing unit calculates positional relations between the white lines and the vehicle, curvatures of the white lines, and a lateral movement speed of the vehicle. Waves transmitted/received by a plurality of wave sensors are controlled by a wave sensor processing unit, and a rear and right/left-side vehicle detection processing unit analyzes reception data of the wave sensors controlled by the wave sensor processing unit and controls a detection level to detect the presence/absence of obstacles on the basis of the presence/absence of white lines, the positional relations between white lines and the vehicle, the curvatures of white lines, and the lateral movement speed of the vehicle.
摘要翻译：要解决的问题为了解决对车辆的驾驶员的死区监视车辆的后方左右两侧的传统装置在检测到另一车辆时警告驾驶员的问题，在某些情况下，错误地检测到护栏，墙面，人行道上的行人等物体车辆，并且任何错误的检测抑制方法都有一些缺点。解决方案：白线检测处理单元分析由车载摄像机捕获的图像，以检测用户的车辆的左侧和右侧的白线划分通道。白线几何位置计算处理单元计算白线与车辆之间的位置关系，白线的曲率以及车辆的横向移动速度。由多个波传感器发送/接收的波由波传感器处理单元控制，后左右侧车辆检测处理单元分析由波传感器处理单元控制的波传感器的接收数据，并控制检测基于白线的存在/不存在，白线与车辆之间的位置关系，白线的曲率以及车辆的横向移动速度来检测障碍物的存在/不存在。版权所有（C）2013，JPO＆INPIT

29. 发明专利

JP2012189441A Object detector 审中-公开
标题翻译：对象检测器
公开(公告)号：JP2012189441A
公开(公告)日：2012-10-04
申请号：JP2011053060
申请日：2011-03-10
申请人： Mitsubishi Electric Corp , 三菱電機株式会社
发明人： NIIYAMA MARIKA , HIRANO TAKASHI
IPC分类号： G01S15/93
摘要： PROBLEM TO BE SOLVED: To provide an object detector that is capable of distinguishing between humans at rest and objects at rest excluding humans, using characteristics of humans.SOLUTION: The object detector includes an ultrasound sensor 1001 that sends and receives ultrasonic waves, a transmission controlling section 1002 that generates a driving signal which drives the ultrasound sensor 1001, and a human distinction processing section 1004 that distinguishes between humans or objects excluding humans on the basis of a reception signal output from the ultrasound sensor 1001. The human distinction processing section 1004 determines that an obstacle is a human on the basis of variation with time of a time series pattern of reception signals obtained through each of the sent and received ultrasonic waves.
摘要翻译：要解决的问题：提供一种能够使用人的特征来区分休息时的人和休息的人物，而不包括人类的物体的物体检测器。解决方案：物体检测器包括发送和接收超声波的超声波传感器1001，产生驱动超声波传感器1001的驱动信号的发送控制部分1002以及区分人或物体的人类区分处理部分1004 基于从超声波传感器1001输出的接收信号排除人。人类区分处理部1004基于通过每个发送的接收信号获得的接收信号的时间序列模式随时间的变化来确定障碍物是人并接收超声波。版权所有（C）2013，JPO＆INPIT

30. 发明专利

JP2009181189A Document check rule creation support device 审中-公开
标题翻译：文件检查规则创建支持设备
公开(公告)号：JP2009181189A
公开(公告)日：2009-08-13
申请号：JP2008017748
申请日：2008-01-29
申请人： Mitsubishi Electric Corp , 三菱電機株式会社
发明人： TANIGAKI KOICHI , HIRANO TAKASHI , OKADA YASUHIRO
IPC分类号： G06F17/21
摘要： PROBLEM TO BE SOLVED: To solve the problem that when creating a rule for checking a prepared document, it is complicated to prepare temporal check conditions, collate the check conditions with a predetermined document, and repeat corrections to obtain desired precision. SOLUTION: This document check rule creation support device is provided with: a check rule storage part storing in advance a check rule obtained by combining check conditions to be used for collation with a document and a message to be presented to a document creator when the check conditions matches the document; a check rule retrieval input part for acquiring retrieval conditions for retrieving the check rule; a check rule discrimination part for discriminating the check rules in the check rule storage part according to whether the check rules satisfies the retrieval conditions from the check rule retrieval input part; and a partial structure extraction part for extracting the partial structure of the check conditions from the discriminated check rules. COPYRIGHT: (C)2009,JPO&INPIT
摘要翻译：要解决的问题：为了解决在制作用于检查准备好的文件的规则的情况下，准备时间检查条件，将检查条件与规定文件进行整理，重复修正以获得期望的精度是复杂的。该文件检查规则创建支持装置具有：检查规则存储部件，预先存储将用于与文档进行对照的检查条件和要呈现给文档创建者的消息相结合的检查条件获得的检查规则当检查条件与文档匹配时; 检查规则检索输入部分，用于获取检索规则的检索条件; 检查规则鉴别部分，用于根据检查规则是否满足来自检查规则检索输入部分的检索条件来鉴别检验规则存储部分中的检查规则; 以及部分结构提取部分，用于从所识别的检查规则中提取检查条件的部分结构。版权所有（C）2009，JPO＆INPIT

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