-
公开(公告)号：KR1020020031015A
公开(公告)日：2002-04-26
申请号：KR1020000062137
申请日：2000-10-21
Applicant: 한국전자통신연구원
IPC: G06T5/40
CPC classification number: G06F17/30259 , H04N19/98
Abstract: PURPOSE: A method for calculating non-linear quantization and a similarity of an edge histogram bin is provided to use a non-linear quantizer, and to encode a histogram bin value generated by 5 directional edge information, so as to reduce a bit amount necessary for storing and improve a similarity calculation speed for search. CONSTITUTION: Images are divided into 4x4 areas(S201). A directional edge histogram is generated(S203). A non-linear quantization table is calculated(S205). The calculated table is stored in a database(S207). An index is normalized(S209). A global and a semi-global histograms are generated(S211). And data are matched(S213). When generating the directional edges, features are extracted by the unit of block for the images divided into 4x4 areas, that is, 16 areas thereby generating 5 edges. The images are divided into sub-images that are not overlapped, and the edge histogram for each sub-image is generated as in the step of (S203).
Abstract translation: 目的：提供一种用于计算非线性量化的方法和边缘直方图bin的相似度，以使用非线性量化器，并且对由5个方向边缘信息生成的直方图bin值进行编码，以便减少必要的比特量 用于存储和提高搜索的相似度计算速度。 构成：图像分为4x4区域（S201）。 生成方向边缘直方图（S203）。 计算非线性量化表（S205）。 将所计算的表存储在数据库中（S207）。 索引被归一化（S209）。 生成全局和半全局直方图（S211）。 数据匹配（S213）。 当产生方向边缘时，通过块的单位提取特征，用于分成4×4区域的图像，即16个区域，从而产生5个边缘。 图像被划分为不重叠的子图像，并且如（S203）的步骤中生成每个子图像的边缘直方图。
-
公开(公告)号：KR100314654B1
公开(公告)日：2001-11-15
申请号：KR1019990008383
申请日：1999-03-12
Applicant: 한국전자통신연구원
IPC: G06T5/40
Abstract: 본발명은 JPEG, MPEG-1, 2의표준을따르는압축된영상데이터로부터수퍼픽셀단위선형양자화를이용하여컬러, 밝기, 그리고에지성분을갖는히스토그램을생성하는방법에관한것이다. 기존의히스토그램생성방법은히스토그램의빈과빈 사이에서발생하는양자화오차의영향에의해정확한특징을반영할수 없었다. 또한, 무채색과유채색을구분하는방법에있어서도경계영역에서오차가발생하였다. 또한, 여러종류의특징정보를동시에사용할경우에는정규화시키는기준이없기때문에다양한특징정보를반영하기가어려웠다. 본발명에서는히스토그램을생성하는기본단위를픽셀들의집합인수퍼픽셀단위로수행하여영상내색상과밝기특징뿐만아니라에지성분의분포까지나타낼수 있도록한다. 또한, 히스토그램의빈과빈 사이에선형적가중치를적용함으로써실수영역에서의연산을통한히스토그램카운터를구현하여히스토그램생성시발생하는오차를감소시킨다. 그리고, 컬러와밝기, 에지에대한특징정보를영상의내용에따라정규화된하나의히스토그램상에표현하여다양한특징정보를반영한다.
-
公开(公告)号：KR1020010002386A
公开(公告)日：2001-01-15
申请号：KR1019990022155
申请日：1999-06-15
Applicant: 한국전자통신연구원
IPC: G06F17/30
Abstract: PURPOSE: A method for building and searching an image DB(database) is provided to manage an image DB more efficiently through a unit of an image data object and to search the DB differently depending on characteristics of each data. CONSTITUTION: A new image data object(22) is inputted to be divided into keyword index data, color index data, shape index data and DB inserting data. According to the keyword index data, non-normalized data is indexed to describe the input image. In case of the color index data, color data are sampled and indexed from the input image data object. In the shape index data, shape data are sampled from the input image data object. According to the DB inserting data, normalized data are sampled from the input image data object. Finally, the divided objects are saved and linked each other.
Abstract translation: 目的：提供一种用于构建和搜索图像DB（数据库）的方法，以通过图像数据对象的单元更有效地管理图像DB，并且根据每个数据的特性不同地搜索DB。 构成：输入新的图像数据对象（22），分成关键字索引数据，颜色索引数据，形状索引数据和DB插入数据。 根据关键字索引数据，对非归一化数据进行索引以描述输入图像。 在颜色索引数据的情况下，从输入图像数据对象对颜色数据进行采样和索引。 在形状索引数据中，从输入图像数据对象中取样形状数据。 根据DB插入数据，从输入图像数据对象取样归一化数据。 最后，分割的对象被保存并相互链接。
-
公开(公告)号：KR1020000037704A
公开(公告)日：2000-07-05
申请号：KR1019980052353
申请日：1998-12-01
Applicant: 한국전자통신연구원
IPC: G06K9/74
Abstract: PURPOSE: An apparatus for inspecting other material on the surface of printed matter using a mask calculation and the method thereof are provided to check the state of cleanliness on the surface of printed matter visually by detecting other material existing on printed matter printed linearly using a mask calculation. CONSTITUTION: An image input unit(10) acquires the image of printed matter(1) to be inspected. A reference point detection and location compensation unit(110) carries out the location compensation after detecting a reference point for the image of the printed matter. An image noise removing unit(120) removes the noise of low frequency for the location-compensated printed matter image. A bit unit OR calculation unit(130) removes printed entities from the printed matter image. A binarization unit(140) separates other material from the background. A connected element extraction and characteristic value estimation unit(150) extracts domains having an illumination of '0' by a connected element and estimates the characteristic value of the connected element.
Abstract translation: 目的：通过使用掩模计算来检查印刷品表面上的其他材料的装置及其方法，用于通过检测使用掩模线性印刷的印刷品上存在的其它材料来目视检查印刷品表面的清洁度的状态 计算。 构成：图像输入单元（10）获取被检查的印刷物（1）的图像。 参考点检测和位置补偿单元（110）在检测到印刷物的图像的参考点之后执行位置补偿。 图像噪声去除单元（120）去除位置补偿印刷物图像的低频噪声。 位单元OR计算单元（130）从打印物图像中去除打印实体。 二值化单元（140）将其他材料与背景分离。 连接元素提取和特征值估计单元（150）通过连接的元素提取具有“0”的照明的域，并且估计所连接的元素的特征值。
-
公开(公告)号：KR1020000037699A
公开(公告)日：2000-07-05
申请号：KR1019980052348
申请日：1998-12-01
Applicant: 한국전자통신연구원
IPC: G06K9/00
Abstract: PURPOSE: A method for extraction of letter lines is provided to extract the location of letter line that is a part of optional image or form document's optional location or territory. CONSTITUTION: A method for extraction of letter lines is composed of search, dilation, calculation, reduction, and generation. An image with letter lines and the image's directional edge is searched(111). According to the searched directional edge, the dots(112) are distinguished into horizontal, vertical, directions(A) and diagonal direction(B). Dilation and calculations(D(A), D(B)) are performed on dots located in horizontal, vertical, and diagonal directions(113). Logical multiple calculation (£D(A)|&£D(B)|) is performed on the previous calculation results and associated dot image is created(114). Associated dot image is reduced, its noises are cleared(115), and block and image coordinates recovery stage is processed to generate letter lines coordinates.
Abstract translation: 目的：提取字母行的提取方法来提取作为可选图像或表单文档可选位置或区域的一部分的字母行的位置。 规定：提取字母行的方法由搜索，扩张，计算，缩小和生成组成。 搜索具有字母线和图像的方向边缘的图像（111）。 根据搜索的方向边缘，点（112）被区分为水平，垂直，方向（A）和对角线方向（B）。 在位于水平，垂直和对角线方向（113）的点上执行扩展和计算（D（A），D（B））。 对先前的计算结果执行逻辑多重计算（£D（A）|＆£D（B）|），并创建相关联的点图像（114）。 关联点图像被减少，其噪声被清除（115），并且处理块和图像坐标恢复阶段以生成字母坐标。
-
Patent Agency Ranking
公开(公告)号：KR1020000037588A
公开(公告)日：2000-07-05
申请号：KR1019980052215
申请日：1998-12-01
Applicant: 한국전자통신연구원
IPC: H04N5/243
Abstract: PURPOSE: An apparatus for removing ghost from image and method thereof are provided to regard dark area of input image as a low image noise and to remove the noise by high pass filtering. CONSTITUTION: An apparatus for removing ghost from image includes an image input(3)and a high pass filter. The image input(3) illuminates light(2) on a metal foil(1) to input the image by the light. The apparatus according to the present invention regards dark area of input image as a low image noise and removes the low image noise by using the high pass filter. The high pass filter further includes a low pass filter(12), a discriminator, and a normalizer. The low pass filter(12) outputs low band noise from the original image. The discriminator is used to acquire the discriminated image from the low band image and the original image.
Abstract translation: 目的：提供一种从图像中去除重影的装置及其方法，以将输入图像的暗区视为低图像噪声，并通过高通滤波消除噪声。 构成：从图像中去除重影的装置包括图像输入（3）和高通滤波器。 图像输入（3）照亮金属箔（1）上的光（2），以通过光输入图像。 根据本发明的装置将输入图像的暗区作为低图像噪声，并且通过使用高通滤波器来去除低图像噪声。 高通滤波器还包括低通滤波器（12），鉴别器和归一化器。 低通滤波器（12）从原始图像输出低频带噪声。 鉴别器用于从低频带图像和原始图像中获取鉴别图像。
-
公开(公告)号：KR100249807B1
公开(公告)日：2000-03-15
申请号：KR1019970067561
申请日：1997-12-10
Applicant: 한국전자통신연구원
IPC: H01L21/66
Abstract: 본 발명은 센서신호로 IC 리드프레임을 확인하여 조명을 구동시키고 동시에 CCD 카메라를 구동시켜 영상을 획득하고 상기 획득된 영상으로부터 리드프레임에 부착된 테이프를 자동검사하는 방법에 관한 것이다.
본 발명은 상기 영상획득부분과 테이프 검사부분을 두 개의 독립된 스레드(thread)로 설계하여 검사가 진행중이더라도 영상을 획득할 수 있는 방법을 제공하며, 특히 영상획득부분은 두 개 이상의 영상 버퍼(buffer)를 사용하여 검사가 진행중이더라도 한 개 이상의 영상을 미리 저장 할 수 있도록하여 전체 검사속도를 높이는 방법과, 리드프레임 패드위치 찾기를 수행한 다음 상기 패드위치로부터 패드의 중심점과 패드의 기울기를 이용하여 기준좌표계를 설정하고 상기 기준좌표계 상에서 테이프의 도면 데이타를 적용하여 테이프 검사를 수행함으로서, 상기 획득된 영상에 나타나는 리드프레임의 위치가, 이송장치의 이송오차나 기계적인 진동에의해 가변적이더라도 정확한 테이프의 위치에서 테이프 검사를 수행하는 방법과, 테이프의 경계선상에 위� �하는 영상정보만을 이용함으로서 고속으로 테이프 검사를 수행하는 방법을 제공하며 임의의 다각형을 확대 및 축소하는 방법을 고안하여 어떤 형태의 테이프에 대해서도 유연하고 쉽게 테이프를 검사하는 방법이다.-
-
-
公开(公告)号：KR1020140079563A
公开(公告)日：2014-06-27
申请号：KR1020120147067
申请日：2012-12-17
Applicant: 한국전자통신연구원
CPC classification number: G06Q10/0635 , G06Q50/01
Abstract: An apparatus for managing risk, based on prediction on social web content, according to the present invention, comprises: a risk vocabulary management unit which extracts and manages vocabulary to be managed as pertaining to risk from social web content; a risk issue prediction analysis quality extraction unit which extracts meta information related to text that is an object for analysis from an original of the social web content, and performs language analysis and sensitivity analysis; a risk prediction modeling unit which models risk prediction analysis through prediction of a statistical and mechanical learning method based on extracted qualities; a risk detection and notification unit which automatically detects the risk that is recognized based on risk prediction models pre-modeled from the social web content, and automatically notifies the detected risk; a risk situation monitoring unit which monitors in real time, a risk state of a risk object when an alarm is raised with respect to the detected risk; and a risk history management unit which receives user feedback for monitored risk information and manages a record of a terminated risk situation.
Abstract translation: 根据本发明，根据社会网页内容的预测来管理风险的装置包括：风险词汇管理单元，其从社交网页内容提取和管理要管理的词汇; 一个风险问题预测分析质量提取单元，从社交网页内容的原始中提取与作为分析对象的文本相关的元信息，并执行语言分析和灵敏度分析; 一种风险预测模型单元，通过基于提取的质量预测统计和机械学习方法来对风险预测分析进行建模; 风险检测和通知单元，其根据从社交网络内容预先建模的风险预测模型自动检测识别的风险，并自动通知检测到的风险; 风险状况监测单元，其实时监测当相对于检测到的风险发生报警时风险对象的风险状态; 以及风险历史管理单元，其接收用于监视的风险信息的用户反馈并管理终止的风险状况的记录。
-
-
-
-
-
-
-
-
-
Search Results
87
records
(took 0.026 seconds)
