公开(公告)号：EP1041379B1

公开(公告)日：2006-08-23

申请号：EP00302745.5

申请日：2000-03-31

申请人： ADE Optical Systems Corporation

发明人： Bills, Richard Earl

IPC分类号： G01N21/95

CPC分类号： G06K9/03 ,  G01N21/9501 ,  G06K9/50

公开(公告)号：EP1622082A2

公开(公告)日：2006-02-01

申请号：EP05016124.9

申请日：2005-07-25

申请人： MITUTOYO CORPORATION

发明人： Tessadro, Ana

IPC分类号： G06T7/00

CPC分类号： G06K9/50 ,  G06K9/4604 ,  G06T7/0004 ,  G06T7/12 ,  G06T7/155 ,  G06T2207/10016 ,  G06T2207/30148 ,  G06T2207/30164

摘要： A method of measuring occluded features in machine vision metrology. The method allows for high precision measurement of features that are partially obscured by an extraneous object (e.g., an overlay grid of a flat panel display screen mask). The method first roughly locates the edge points of the feature to measure. In one embodiment, the rough edge location is performed by using morphology filters to delete the extraneous object (e.g., the grid) and then doing edge detection on the filtered image. Once the rough edge locations are determined, the method then returns to the original unaltered image and for each of the roughly located edge points does edge detection in a close neighborhood of the point in the original unaltered image. These refined edge points are then used to fit the feature for highly precise measurements.

摘要翻译： 一种测量机器视觉计量学中闭塞特征的方法。 该方法允许对由外部物体（例如，平板显示屏幕掩模的覆盖网格）部分遮蔽的特征的高精度测量。 该方法首先粗略地定位要测量的要素的边缘点。 在一个实施例中，通过使用形态滤波器来去除外部物体（例如，网格）并且然后对经过滤的图像进行边缘检测来执行粗略边缘位置。 一旦确定了粗边缘位置，则该方法然后返回到原始未改变的图像，并且对于每个粗略定位的边缘点，在原始未改变图像中的点的邻近区域进行边缘检测。 然后将这些精细的边缘点用于适应高精度测量的特征。

公开(公告)号：EP1564683A1

公开(公告)日：2005-08-17

申请号：EP05100844.9

申请日：2005-02-07

申请人： Xerox Corporation

发明人： Curry, Donald J. ,  Nafarieh, Asghar ,  Kletter, Doron ,  Thayer, Todd W.

IPC分类号： G06T5/00

CPC分类号： G06K9/50 ,  G06T7/11 ,  G06T7/194 ,  G06T2207/10008 ,  G06T2207/30176 ,  H04N1/56

摘要： Image data is analyzed line by line to identify pixels having similar color and edge characteristics. The pixels on one line are linked to pixels on another line, if they share similar characteristics. The linked pixels are then grouped together in regions, and the average color of the region is calculated. An image map is generated that identifies the regions to which the pixels belong, along with a final table of the regions appearing in the image.

公开(公告)号：EP1229497A3

公开(公告)日：2003-05-07

申请号：EP02250445.0

申请日：2002-01-22

申请人： MATSUSHITA ELECTRIC INDUSTRIAL CO., LTD.

发明人： Guo, Jinhong K.

IPC分类号： G06T9/20 ,  G06T5/00

CPC分类号： G06K9/50 ,  G06K9/48 ,  G06K2209/01 ,  G06T9/005 ,  G06T9/20

摘要： Connected components and contour following identifying contours within pixel-based image data. The image data is expressed as a grid of columns and rows. A scan order is established over the grid to define a parent-child relationship between contiguous pixels in adjacent rows and to define a sibling relationship among non-contiguous pixels in the same row. A run data structure is established in computer-readable memory that defines a run member by its row position and by its starting and ending column positions. The run data structure further defining parent, child and sibling structures for storing information about the parent-child relationships and sibling relationships of pixels associated with the run member. Image data is scanned according to the scan order to identify contiguous pixels of a predetermined state as identified run members. Parent-child and sibling relationships of the identified run members are determined. The run data structure is populated with the row position and starting and ending column positions of the identified run member and with the parent-child and sibling relationships of the identified run member. The populated run data structure is used to traverse the parent-child and sibling relationships and thereby identify contours within the pixel-based image data.

公开(公告)号：EP1041379A2

公开(公告)日：2000-10-04

申请号：EP00302745.5

申请日：2000-03-31

申请人： ADE Optical Systems Corporation

发明人： Bills, Richard Earl

IPC分类号： G01N21/88

CPC分类号： G06K9/03 ,  G01N21/9501 ,  G06K9/50

摘要： The present invention provides apparatus and methods for scanning a workpiece for defects with increased resolution and sensitivity relative to conventional workpiece inspection systems. Specifically, the apparatus and methods of the present invention repeatedly scan different portions of the workpiece with an optical beam that forms a scan region on the surface of the workpiece having in-scan and cross-scan dimensions. The in-scan dimension of the scan region is parallel to the direction with which the optical beam is scanned, and the cross-scan dimension is in a direction that is generally perpendicular to the direction in which the optical beam is scanned. The apparatus and methods of the present invention filter the signal reflected from the surface of the workpiece in the cross-scan direction using a cross-scan filter. As such, the method and apparatus of the present invention effectively remove a significant portion of the noise in the reflected signals used for detecting for defects in the workpiece. Further, in some embodiments, the apparatus and method also filter the reflected signal in the in-scan direction using an in-scan filter. By filtering the reflected signals in either a cross-scan or both a cross-scan and in-scan directions, the apparatus and methods of the present invention remove noise from the reflected signals, thereby increasing the sensitivity and resolution of the workpiece inspection device.

摘要翻译： 本发明提供的装置和用于扫描用于与分辨率和敏感度相对于常规的工件检查系统增加的缺陷的工件的方法。 具体地，该装置和本发明的方法反复扫描与在光束形式那样具有扫描和交叉扫描尺寸的工件的表面上的扫描区域内的工件的不同部分。 扫描区域的在扫描尺寸平行于与该光束被扫描的方向，和交叉扫描尺寸在一个方向上所做的是垂直于光束扫描方向基因的反弹。 的装置和本发明过滤器从使用交叉扫描滤波器在交叉扫描方向的工件的表面反射的信号的方法。 如测试，本发明的方法和装置有效地用于检测在工件缺陷的反射信号中去除噪声的显著部分。 此外，在一些实施例中，该装置和方法因此使用扫描滤波器中扫描方向滤波器中的反射信号。 通过在任何一个交叉扫描或两个交叉扫描和在扫描方向中的滤波的反射信号，该装置和本发明的方法从反射的信号中去除噪声，从而增加了工件检查装置的灵敏度和分辨率。

公开(公告)号：EP0965943A2

公开(公告)日：1999-12-22

申请号：EP99304385.0

申请日：1999-06-04

申请人： MATSUSHITA ELECTRIC INDUSTRIAL CO., LTD.

发明人： Miyazaki, Keiichi ,  Minami, Masato ,  Kouda, Toshiyuki

IPC分类号： G06K9/46 ,  G06K9/20

CPC分类号： G06K9/00449 ,  G06K9/48 ,  G06K9/50 ,  G06K2209/01

摘要： A method of recognizing characters of a document including ruled lines. After separating the characters from the ruled lines, a bounding box and a corresponding character pattern are extracted for each character (hereinafter, "the character"). Contour information and ground (white pel area) information are extracted as fundamental features from the character pattern. Various statistical features are extracted based on the fundamental features. Various structural features are extracted based on the fundamental features and the character pattern. On the basis of the extracted statistical features, some candidates for the character and corresponding degrees of similarity are found to provide at least such one of the candidates as has largest degree(s) of similarity and degree(s) of similarity associated with the provided candidate(s). A final candidate for the character pattern is output based on the provided candidate(s), the provided degree(s) of similarity, the structural features and the bounding box.

摘要翻译： 认识到一个文件包括分格线的字符的方法。 分离的分格线中的字符之后，边界框和相应的字符图形被提取的每个字符（在后，“字符”）。 轮廓信息和地（白色像素区域）信息作为从根本上提取从字符图案特征。 各种统计特征是基于基本特征提取。 各种结构特征是基于基本的功能和字符图案提取。 所提取的统计特征的基础上，对角色和相应的相似度有些考生被发现提供至少检查候选人之一，因为有大量最大的与所提供的相关联的相似的相似性和度（S）的程度（S） 候选者。 为字符图案A最终候选是基于所提供的候选者输出，相似的程度提供（一个或多个）时，结构特征和边界框。

公开(公告)号：EP0896295A3

公开(公告)日：1999-11-03

申请号：EP98305746.4

申请日：1998-07-17

申请人： XEROX CORPORATION

发明人： Handley, John C. ,  Eschbach, Reiner

IPC分类号： G06K9/46

CPC分类号： G06K9/50 ,  G06K2209/01

摘要： A method and system for performing optical character recognition wherein data remains in compressed form. Character features are extracted in the spatial domain, and are distinguished from recognition features in the compressed domain wherein said character image is represented in compact form. OCR is conducted within a compressed data stream wherein a document is scanned; said document is compressed; recognition feature extraction is performed on said document; and recognition features extracted from said document are classified.

公开(公告)号：EP0855680A1

公开(公告)日：1998-07-29

申请号：EP98200191.9

申请日：1998-01-23

申请人： Océ-Technologies B.V.

发明人： Luttmer, Maurice Lambert Martin

IPC分类号： G06T9/00

CPC分类号： G06T9/005 ,  G06K9/50

摘要： A method of performing an image processing operation directly on the runs of a runlength encoded bitmap of pixels. For this purpose, the method comprises a first selection step for selecting in the first instance at least two runs from one or more series of runs of one or more images on the basis of pixel position information. A runlength processing operation is then performed on the selected runs in accordance with the image processing operation in order to obtain a resultant run having a resultant runlength and a resultant pixel value. In a shortening step the runlengths of the processed runs are shortened by a value derived from the determined resultant runlength. In a step size determining step a step size is derived from the resultant runlength whereafter in a second selection step one or more shortened runs and/or one or more following runs from the one or more series of runs of the one or more images are selected on the basis of the step size information and the above-mentioned steps are performed again until all the runs have been processed.

摘要翻译： 对像素的游程编码位图的运行直接执行图像处理操作的方法。 为此，该方法包括第一选择步骤，用于根据像素位置信息从一个或多个图像的一行或多行中选择至少两次。 然后根据图像处理操作对所选择的行进行游程长度处理操作，以获得具有合成游程长度和合成像素值的合成运行。 在缩短步骤中，经处理的运行的运行长度被缩短从确定的合成运行长度导出的值。 在步长确定步骤中，从所得到的游程长度导出步长，然后在第二选择步骤中，选择一个或多个缩短的运行和/或从一个或多个图像的一个或多个运行序列的一个或多个跟随运行 基于步长信息，再次执行上述步骤，直到所有的运行都被处理。

公开(公告)号：EP0801362A2

公开(公告)日：1997-10-15

申请号：EP97200936.9

申请日：1997-04-01

申请人： Océ-Technologies B.V.

发明人： Luttmer, Maurice Lambert Martin

IPC分类号： G06T9/00

CPC分类号： G06K9/50 ,  G06T9/005

摘要： The invention is related to a method for converting a series of drawing instructions suitable for assigning pixel values corresponding to said drawing instructions to pixels positioned in an image arranged in rows and columns of pixels, to a series of runlengths and corresponding pixel values. A runlength hereby represents the number of successive pixels with mutual identical pixel values while the series of runlengths and pixel values is arranged according to a first order as determined by successive pixel positions. First the runlengths of the pixel values as generated by the drawing instructions are determined in a second order which is different from the first order. Consequently, the insert positions of said determined runlengths are determined on the basis of pixel positions which are corresponding with said determined runlengths. On the basis of these insert positions, said deteremined runlengths are inserted in the series of runlengths according to an prescription for inserting whereby runlengths in the series which have positions in common with said runlengths to be inserted, are modified in dependance on the said runlengths and pixel values to be inserted.

摘要翻译： 本发明涉及一种用于将适合于将对应于所述绘制指令的像素值分配给一系列绘制指令的一系列绘制指令转换为一系列游程长度和对应的像素值的方法。 因此，游程长度表示具有相互相同像素值的连续像素的数量，同时根据由连续的像素位置确定的一阶排列一系列游程长度和像素值。 首先，以与第一顺序不同的第二顺序来确定由绘图指令产生的像素值的游程长度。 因此，基于与所确定的游程长度对应的像素位置来确定所确定的游程长度的插入位置。 根据这些插入位置，根据用于插入的处方，所述确定的游程长度以一系列的游程长度被插入，其中具有与要插入的游程长度相同的位置的系列中的游程长度依赖于所述游程长度被修改， 要插入的像素值。

公开(公告)号：EP0635981A1

公开(公告)日：1995-01-25

申请号：EP94304935.3

申请日：1994-07-05

申请人： SHARP KABUSHIKI KAISHA

发明人： Yoshida, Shigeo

IPC分类号： H04N7/14

CPC分类号： H04N7/148 ,  G06K9/00234 ,  G06K9/342 ,  G06K9/50 ,  G06T7/11 ,  G06T7/90 ,  G06T2207/10016 ,  G06T2207/10024 ,  G06T2207/30201 ,  H04N19/20

摘要： The present invention is intended to exactly extract a feature-region along its contour line without erroneously extracting any moving object other than the feature-region. An extraction signal from a feature-region main-portion extracting circuit(1) is sent to a small region eliminating circuit (2) whereby an erroneously extracted small region is detected and eliminated as an erroneously extracted noisy portion. The small region eliminating circuit(2) compares a small region with four detection patterns, discriminates it as a region other than the feature face-region if no match is found and causes a center pixel of a window of 3 × 3 pixels to have a logical value of zero. An output of the small region eliminating circuit (2) is sent to a blank filling circuit (3) which compares its input with four detection patterns, discriminates it as a face-region if a match is found and makes a central pixel of a window of 13 × 13 pixels have a logical value of 1. The above-mentioned extraction method can reliably extract features of the face-region and is free from the erroneous extraction of any moving object other than the face.

摘要翻译： 本发明旨在沿着其轮廓线精确地提取特征区域，而不会错误地提取除特征区域之外的任何移动物体。 来自特征区域主要部分提取电路（1）的提取信号被发送到小区域消除电路（2），由此错误地提取的小区域被检测和消除，作为错误地提取的噪声部分。 小区域消除电路（2）将小区域与四个检测图案进行比较，如果不匹配，则将其区分为特征面部区域以外的区域，并使3×3像素的窗口的中心像素具有 逻辑值为零。 将小区域消除电路（2）的输出发送到空白填充电路（3），其将其输入与四个检测图案进行比较，将其识别为面部区域，如果发现匹配并使窗口的中心像素 13×13像素的逻辑值为1.上述提取方法可以可靠地提取面部区域的特征，并且没有错误地提取除脸部之外的任何移动物体。