会员体验
专利管家(专利管理)
工作空间(专利管理)
风险监控(情报监控)
数据分析(专利分析)
侵权分析(诉讼无效)
联系我们
交流群
官方交流:
QQ群: 891211   
微信请扫码    >>>
现在联系顾问~
热词
    • 11. 发明专利
    • Pattern recognizing system
    • 模式识别系统
    • JPS5962984A
    • 1984-04-10
    • JP17085882
    • 1982-10-01
    • Hitachi Ltd
    • KAGEYAMA SAIJIENDOU HIROHIDENAKANO YASUAKIOKADA KUNIHIROYAMAMOTO MASAO
    • G06K9/00G06K9/03
    • G06K9/00
    • PURPOSE:To recognize patterns accurately by controlling the color and the luminance, with which patterns or categories are displayed, in accordance with preliminarily measured parameters with respect to the first candidate pattern giving an answer category. CONSTITUTION:An input pattern 40a is inputted to an observing part 41, and the observing part 41 converts the input pattern 40a to an electric signal 41a. This electric signal 41a is subjected to preprocessing or/and segmenting in a signal processing part 42 and is conveted to a unit pattern 42a. The pattern 42a is inputted to a recognizing part 43, and the resemblance between the unit pattern 42a and each standard pattern 44a stored in an incorporating part of a dictionary part 44 is evaluated in the recognizing part 43. On the evaluation value, an answer category 43a is outputted or this output is rejected. A signal 43a or 43b resulting from these recognition is outputted to a display part 45.
    • 目的:通过根据给出答案类别的第一候选图案的初步测量参数,通过控制显示图案或类别的颜色和亮度来准确地识别图案。 构成:将输入图案40a输入到观察部41,观察部41将输入图案40a转换为电信号41a。 该电信号41a在信号处理部分42中经受预处理或/和分段,并且被转换为单元图案42a。 图案42a被输入到识别部分43,并且在识别部分43中评估存储在字典部分44的合并部分中的单元图案42a和每个标准图案44a之间的相似度。在评估值上,应答类别 43a被输出或该输出被拒绝。 由这些识别产生的信号43a或43b被输出到显示部分45。
    • 16. 发明专利
    • PICTURE FILE DEVICE
    • JPS60114967A
    • 1985-06-21
    • JP22202583
    • 1983-11-28
    • HITACHI LTD
    • KUNISAKI OSAMUENDOU HIROHIDE
    • G06K9/00G06F17/30G06T1/00G09G1/00G09G5/00G09G5/40
    • PURPOSE:To attain picture filing by coding automatically a key word existing at an optional position in a document descibedon a document through character recognition and adding the keyword automatically to an optional form. CONSTITUTION:A character string of an underlined part of picture information processed by a picture processing section 23 is extracted one by one character by a recognition object character extracting section 4. When a red underline is detected, the character for the part is extracted and read by using a means ANDing an underline detection signal and a character detection signal. A character recognizing section 25 recognizes an extracted character string one by one and converts it into an internal code. A code train 251 and picture processed picture information 252 obtained in this way are inputted to a file production processing section 26. The file production processing section 26 adds a retrieval keyword code 251 to the picture processed picture information 252 in a prescribed format to form a picture file data 27.
    • 17. 发明专利
    • PRODUCTION OF STANDARD PATTERN
    • JPS6033680A
    • 1985-02-21
    • JP14104283
    • 1983-08-03
    • HITACHI LTD
    • NAKANO YASUAKIENDOU HIROHIDEKUNISAKI OSAMUOKADA KUNIHIROOZAKI TOSHITSUGU
    • G06K9/68
    • PURPOSE:To obtain a dictionary for recognition consisting of groups of the standard pattern optimum to each user by storing the information showing the change of category names of the standard patterns into said dictionary. CONSTITUTION:A feature pattern of input character is compared with a standard pattern of a memory 9 for recognition at a recognizing part 8. For the loading operation to a memory 9, a personal information dictionary is loaded to a memory 11 from an external memory 10 at first. The personal information dictionary contains a flag which decides whether this dictionary is used for each standard pattern, a flag which decides whether the category name is rewritten, and the information showing the category name of the other side if the category name is rewritten. When the loading is through with the personal dictionary, the standard patterns of a recognition dictionary are loaded successively. Each standard pattern is loaded to the memory 9 in case the pattern is used with reference to the corresponding flag of the personal dictionary and also used with rewriting of the category name. In the case of the latter, the category name is changed and loaded. The loading operation are through when the above-mentioned operation is carried out with all standard patterns.
    • 18. 发明专利
    • Pattern recognition system
    • 模式识别系统
    • JPS59148984A
    • 1984-08-25
    • JP2274683
    • 1983-02-16
    • Hitachi Ltd
    • OKADA KUNIHIROENDOU HIROHIDENAKANO YASUAKIKUNISAKI OSAMUKUROSU YASUO
    • G06K9/03
    • PURPOSE: To prevent the continuation of rejection and reading error by measuring the rejection ratio of one business form and switching a deciding parameter.
      CONSTITUTION: An area including a character pattern drawn in an input business form 1 is converted into a binary pattern by a photoelectric conversion part 2 and the binary pattern is supplied to a preprocessing part 3. The preprocessing part 3 segments only one character from the business form image expressed as the binary pattern and sends the segmented character to a buffer memory 4. The character pattern stored in the buffer memory 4 is processed by a feature extracting part 5 and the extracted characteristic variable is stored in a buffer memory 6. On the other hand, a category name and a characteristic variable for a previously fixed standard pattern are found out and stored in the dictionary 7. The contents stored in the buffer memory 6 are successively collated with the contents read out from the dictionary 7 and the collated result is sent to a discrimination part 9. The output of the deciding part 9 is sent to a post-processing part 10 and the insertion of the rejected character or the like is executed.
      COPYRIGHT: (C)1984,JPO&Japio
    • 目的:通过测量一个业务表单的拒绝率和切换决定参数来防止拒绝和读取错误的继续。 构成:通过光电转换部分2将包含以输入商业形式1绘制的字符图案的区域转换为二进制图案,并将二进制图案提供给预处理部分3.预处理部分3仅从业务分割出一个字符 形式图像表示为二进制图案,并将分割的字符发送到缓冲存储器4.存储在缓冲存储器4中的字符模式由特征提取部分5处理,并且提取的特征变量存储在缓冲存储器6中。 另一方面,找到用于先前固定的标准模式的类别名称和特征变量并将其存储在字典7.存储在缓冲存储器6中的内容与从字典7读出的内容和对照结果连续对照 被发送到鉴别部分9.决定部分9的输出被发送到后处理部分10,并且插入被拒绝的特征 呃等。
    • 19. 发明专利
    • Optical character reader
    • 光学读写器
    • JPS59784A
    • 1984-01-05
    • JP10973682
    • 1982-06-28
    • Hitachi Ltd
    • OKADA KUNIHIROENDOU HIROHIDEITOU SATOSHITAKAHASHI TSUGIO
    • G06K9/20
    • G06K9/20
    • PURPOSE:To use a slip with free format without reference mark, by overlapping the range of the photoelectric conversion of a one-dimension photoelectric converter for a size for one character's share or a one line's share of pitch or over. CONSTITUTION:The slip is stopped with a paper feed mechanism when the slip is moved toward the Y direction by a distance for one line or one character or over less than the range read with a sensor in the photoelectric conversion section 50. The image converted into an electric signal with the sensor is converted into a digital signal at an A/D converter 51 and stored in a video memory 52. A character recognizing section 53 reads out the image of the video memory 52 for attaining character recognition.
    • 目的:使用无参考标记的滑块,通过将一维光电转换器的光电转换范围重叠为一个字符的份额或一行的音高或以上的份额。 构成:当滑动向Y方向移动一行或一个字符的距离或小于光电转换部分50中用传感器读取的范围时,使用进纸机构停止滑动。转换成 具有传感器的电信号在A / D转换器51处被转换成数字信号并存储在视频存储器52中。字符识别部分53读出用于获得字符识别的视频存储器52的图像。
    • 20. 发明专利
    • PHOTOELECTRIC TRANSDUCER
    • JPS5851385A
    • 1983-03-26
    • JP14946081
    • 1981-09-24
    • HITACHI LTDHITACHI KOKI KK
    • OKADA KUNIHIROENDOU HIROHIDESAITOU SUSUMU
    • G02B26/10G03B42/02G06K9/20G06T1/00H04N1/028H04N5/257
    • PURPOSE:To obtain efficiently the output having no variance, by attaching photoelectric transducers, which are long in the direction orthogonal to the movement direction of an object face, near the moving object face and scanning a light spot in the longitudinal direction of photoelectric transducers and converting photoelectrically reflected lights. CONSTITUTION:The laser light from a laser light source 5 is condensed by a lens 9 to irradiate a rotary polyhedral mirror 6. The laser light reflected by the rotary polyhedral mirror moves in the x-axis direction to scan an object 4 moving in the y-axis direction, and the optical path length difference generated in this case is corrected by a correcting lens such as an Ftheta lens. Photoelectric transducers 81 and 82 which are long in the x-axis direction are arranged near the object 4. Photoelectric conversion faces 811 and 821 are coupled with respective anodes 812 and 822 to receive the reflected light from the object 4. Thus, a high-resolution output having no variance is obtained efficiently with a simple device. If the device is stored in a case, to which the external light is not incident, to improves S/N or a light scattering medium which prevents the abnormal output due to the specular reflection is used, the better result is obtained.