-
公开(公告)号：US09449222B2
公开(公告)日：2016-09-20
申请号：US14674882
申请日：2015-03-31
Applicant: FUJITSU LIMITED
CPC classification number: G06K9/00375 , G06T7/11 , G06T7/12 , G06T7/136 , G06T7/194 , G06T2207/10024
Abstract: The present disclosure relates to an image processing device and method. The device includes: an acquiring unit for acquiring an image containing an object image; a calculating unit for calculating one or more candidate object image areas based on color and edge characteristics of the image; a selecting unit for selecting a candidate object image area with the maximal similarity to a predetermined color model as a first object image area from the one or more candidate object image areas; a determining unit for determining other of the one or more candidate object image areas as a non-object image area or a second object image area with the first object image area; a creating unit for creating an adaptive color model; and a detecting unit for detecting, based on the adaptive color model, an area containing the object image in the each of the first and second object image areas.
Abstract translation: 本公开涉及一种图像处理装置和方法。 该装置包括：获取单元，用于获取包含对象图像的图像; 计算单元，用于基于图像的颜色和边缘特性来计算一个或多个候选对象图像区域; 选择单元，用于从所述一个或多个候选对象图像区域中选择与预定颜色模型最大相似度的候选对象图像区域作为第一对象图像区域; 确定单元，用于将所述一个或多个候选对象图像区域中的另一个确定为具有所述第一对象图像区域的非对象图像区域或第二对象图像区域; 用于创建自适应颜色模型的创建单元; 以及检测单元，用于基于所述自适应颜色模型，检测在所述第一和第二对象图像区域中的每一个中包含所述对象图像的区域。
-
公开(公告)号：US09418307B2
公开(公告)日：2016-08-16
申请号：US14598606
申请日：2015-01-16
Applicant: FUJITSU LIMITED
CPC classification number: G06K9/44 , G06K9/4642 , G06K2209/011 , G06T5/002 , G06T5/005 , G06T5/10 , G06T7/194 , G06T2207/20021 , G06T2207/20064 , G06T2207/30176 , H04N1/4092
Abstract: A method and apparatus for smoothing an image representing foreground or background. The method comprises: dividing the image in a specified granularity; for a block divided containing a blank pixel and a non-blank pixel, modifying a value of the blank pixel to a value obtained from a value of the non-blank pixel to obtain a filled block; obtaining the block smoothed by performing singular value decomposition on a matrix of pixels of the filled block, wherein values in the matrix of pixels of the block smoothed is calculated as a product of the obtained singular value, a left singular vector, and a transpose of a right singular vector; and for each of the non-blank pixels in the block smoothed, restoring the values of the non-blank pixels to initial values before the filling step, so as to obtain a reconstructed block.
Abstract translation: 一种用于平滑代表前景或背景的图像的方法和装置。 该方法包括：以特定粒度划分图像; 对于包含空白像素和非空白像素的块分割，将空白像素的值修改为从非空白像素的值获得的值，以获得填充块; 通过对填充块的像素矩阵执行奇异值分解来获得平滑的块，其中平滑的块的像素矩阵中的值被计算为所获得的奇异值，左奇异矢量和转置 一个正确的奇异矢量 并且对于平滑的块中的每个非空白像素，在填充步骤之前将非空白像素的值恢复为初始值，以便获得重构块。
-
33.发明授权Method and apparatus for correcting corner point of image and image processing device 有权用于校正图像和图像处理装置的角点的方法和装置公开(公告)号：US09092844B2
公开(公告)日：2015-07-28
申请号：US13740571
申请日：2013-01-14
Applicant: Fujitsu Limited
Inventor： Pan Pan , Yuan He , Jun Sun , Satoshi Naoi
CPC classification number: G06T5/001 , G06T3/0031 , G06T5/006 , G06T7/12 , G06T2207/10008 , G06T2207/20164 , G06T2207/30176
Abstract: The present invention discloses a method and apparatus for correcting a corner point of an image and an image processing device. The method includes: determining first candidate corner points of an initial corner point in a first local region; obtaining information related to the image in a second local region; selecting, among the first candidate corner points of the initial corner point, the first candidate corner points meeting a predetermined condition, as second candidate corner points of the initial corner point according to the information; and correcting the initial corner point using the second candidate corner points of the initial corner point. The apparatus is configured to perform the processes of the method. The image processing device includes the apparatus for correcting a corner point of an image. With the technology, a roughly detected corner point can be corrected.
Abstract translation: 本发明公开了一种用于校正图像的角点和图像处理装置的方法和装置。 该方法包括：确定第一局部区域中的初始角点的第一候选角点; 获取与第二局部区域中的图像有关的信息; 在所述初始角点的所述第一候选角点中，选择满足预定条件的所述第一候选角点作为所述信息的所述初始角点的第二候选角点; 以及使用所述初始角点的第二候选角点来校正所述初始角点。 该装置被配置为执行该方法的处理。 图像处理装置包括用于校正图像的角点的装置。 使用该技术，可以校正粗略检测的角点。
-
34.发明授权Device, method, and scanner correcting document image using correcting mesh constructed according to vertical and horizontal boundaries and adjusted extended filtered lines 有权设备，方法和扫描仪使用根据垂直和水平边界构造的校正网格和调整的扩展滤波线来校正文档图像公开(公告)号：US09083909B2
公开(公告)日：2015-07-14
申请号：US14547472
申请日：2014-11-19
Applicant: FUJITSU LIMITED
CPC classification number: G06T7/0085 , G06T3/0031 , G06T5/006 , H04N1/387 , H04N2201/0081 , H04N2201/0434
Abstract: Embodiments of the present invention provide a device and method for correcting a document image and a scanner. Wherein the device includes: an extracting unit configured to extract boundaries in a first direction of the document image and extract lines in the first direction according to a content of the document image, a filtering unit configured to filter the extracted lines, a constructing unit configured to extend and adjust the filtered lines and construct a correcting mesh, and a correcting unit configured to perform correcting according to the correcting mesh. An accurate correcting model can be constructed by taking document boundaries and document contents into account in constructing the correcting model and constructing a correcting mesh after filtering, extending and adjusting the extracted lines, thereby effectively eliminating distortion in the document image.
Abstract translation: 本发明的实施例提供了一种用于校正文档图像和扫描仪的装置和方法。 其中，所述设备包括：提取单元，被配置为根据所述文档图像的内容提取所述文档图像的第一方向上的边界并且提取所述第一方向上的线;滤波单元，被配置为对所提取的线进行滤波;构造单元， 扩展和调整滤波后的线并构造校正网格;以及校正单元，被配置为根据校正网格执行校正。 可以通过在构建校正模型时考虑文档边界和文档内容，并在过滤，扩展和调整提取的线之后构建校正网格，从而有效消除文档图像中的失真，从而构建精确的校正模型。
-
公开(公告)号：US09082181B2
公开(公告)日：2015-07-14
申请号：US13875376
申请日：2013-05-02
Applicant: FUJITSU LIMITED
Inventor： Jun Sun , Yifeng Pan , Satoshi Naoi
CPC classification number: G06T7/004 , G06K9/00442 , G06T7/70
Abstract: An image processing method and apparatus is provided. The image processing method includes steps of: generating a first scale binary image from an image, wherein the first scale is smaller than the original scale of the image; detecting at least one text line in the image based on the first scale binary image; generating a second scale binary image from the image, wherein the second scale is larger than the first scale; for each text line, calculating a similarity between corresponding sections in the first scale binary image and the second scale binary image, and removing the text line for which the similarity is lower than a predetermined level; for one or more of the remaining text line(s), performing OCR on corresponding section(s) in the second scale binary image to determine character orientation(s) of corresponding text line(s); and determining the orientation of the image according to the determined character orientation(s).
Abstract translation: 提供了一种图像处理方法和装置。 该图像处理方法包括以下步骤：从图像生成第一刻度二进制图像，其中第一刻度小于图像的原始尺度; 基于所述第一缩放二进制图像来检测所述图像中的至少一个文本行; 从所述图像生成第二尺度二进制图像，其中所述第二刻度大于所述第一刻度; 对于每个文本行，计算第一缩放二进制图像和第二缩放二进制图像中的相应部分之间的相似度，并且去除相似度低于预定水平的文本行; 对于一个或多个剩余的文本行，在第二比例二进制图像中的相应部分上执行OCR以确定相应文本行的字符取向; 以及根据确定的字符取向来确定图像的取向。
-
Patent Agency Ranking
公开(公告)号：US11586926B2
公开(公告)日：2023-02-21
申请号：US16251471
申请日：2019-01-18
Applicant: FUJITSU LIMITED
Inventor： Liuan Wang , Jun Sun
Abstract: A method and apparatus of accelerating deep learning, and a deep neural network are provided. The method comprises: randomly initializing weights and biases of a deep neural network as n-bit fixed-point numbers; reducing data in a plurality of layers in the deep neural network that have calculation amounts are greater than a first predetermined threshold as m-bit fixed-point numbers, where m and n are integers and m
-
公开(公告)号：US11556824B2
公开(公告)日：2023-01-17
申请号：US16902756
申请日：2020-06-16
Applicant: FUJITSU LIMITED
Inventor： Chaoliang Zhong , Wensheng Xia , Ziqiang Shi , Jun Sun
Abstract: The present disclosure relates to methods for estimating an accuracy and robustness of a model and devices thereof. According to an embodiment of the present disclosure, the method comprises calculating a parameter representing a possibility that a sample in the first dataset appears in the second dataset; calculating an accuracy score of the model with respect to the sample in the first dataset; calculating a weighted accuracy score of the model with respect to the sample in the first dataset, based on the accuracy score, by taking the parameter as a weight; and calculating, as the estimation accuracy of the model with respect to the second dataset, an adjusted accuracy of the model with respect to the first dataset according to the weighted accuracy score.
-
公开(公告)号：US11455525B2
公开(公告)日：2022-09-27
申请号：US16180400
申请日：2018-11-05
Applicant: FUJITSU LIMITED
Abstract: A method and apparatus of open set recognition, and a computer-readable storage medium are disclosed. The method comprises acquiring auxiliary data and training data of known categories for open set recognition, training a neural network alternately using the auxiliary data and the training data, until convergence; extracting a feature of data to be recognized for open set recognition, using the trained neural network; and recognizing a category of data to be recognized, based on the feature of the data to be recognized.
-
公开(公告)号：US11263108B2
公开(公告)日：2022-03-01
申请号：US16847822
申请日：2020-04-14
Applicant: FUJITSU LIMITED
Inventor： Shenbin Zhang , Bingfeng Pi , Jun Sun
Abstract: A device for testing a blockchain network is provided. The device includes a processor. The processor is configured to calculate a matching degree between a configuration of the blockchain network and a template configuration corresponding to application scenario of the blockchain network; calculate performance of the blockchain network; calculate a health degree for the blockchain network; and determine whether the blockchain network is in a stable state based on the matching degree, the performance and the health degree.
-
公开(公告)号：US10990438B2
公开(公告)日：2021-04-27
申请号：US16397553
申请日：2019-04-29
Applicant: FUJITSU LIMITED
Inventor： Shenbin Zhang , Bingfeng Pi , Jun Sun
Abstract: Disclosed are a method and apparatus for managing effectiveness of an information processing task in a decentralized data management system. The method comprising: sending requests for multiple information processing tasks by a client to multiple execution subjects, transmitting information processing tasks in a sequential information processing task list in an order to the multiple execution subjects; caching the requested information processing tasks to a task cache queue, caching the sequential information processing task list as a whole to the task cache queue; judging whether each information processing task in the task cache queue satisfies a predetermined conflict condition; moving the information processing task to a conflict task queue if it is determined that the task satisfies the predetermined conflict condition, deleting the task from the conflict task queue and caching the task to the task cache queue when the predetermined conflict condition is not satisfied.
-
-
-
-
-
-
-
-
-
Search Results
51
records
(took 0.019 seconds)
