-
-
公开(公告)号:KR1020160103225A
公开(公告)日:2016-09-01
申请号:KR1020150025224
申请日:2015-02-23
Applicant: 한국전자통신연구원
IPC: G06T7/20
CPC classification number: G06T7/20 , G06T2207/30201
Abstract: 본발명의실시예에따른멀티스레드를이용한실시간얼굴검출방법은고해상도영상내에서실시간으로얼굴을검출하는방법에있어서, 후처리구간을결정한후 영상데이터를입력받고추적구간을결정하는단계; 특정프레임에대한얼굴검출을처리하는단계; 연이어입력되는프레임에대해상기추적구간만큼프레임을저장하는단계; 상기얼굴검출처리값을통해추적할얼굴개수및 위치를파악하여추적기를생성하는단계; 연이은프레임들에대해후처리가수행되기전까지추적구간동안얼굴검출처리값을이용하여추적기를계속누적생성하여추적하는단계; 상기생성된추적기에대해상기후처리구간동안후처리를수행하는단계; 및사용자움직임을판단하여움직임이많은경우얼굴추적시간을짧게설정하고움직임이적은경우는얼굴추적시간을길게설정해주는단계를포함할수 있다.
Abstract translation: 根据本发明的一个实施例的用于实时地检测面部的方法,涉及一种用于在高分辨率图像中实时检测面部的方法,包括以下步骤:在确定后视图之后接收图像数据, 处理部分,并确定跟踪部分; 处理特定框架的面部检测; 存储与连续输入的帧的跟踪部分一样多的帧; 通过抓取通过面部检测处理值追踪的面部的数量和位置来产生跟踪器; 通过在执行连续帧的后处理之前通过在跟踪部分期间使用面部检测处理值来连续地累积和生成跟踪器来执行跟踪; 在生成的跟踪器的后处理部中进行后处理; 并设置一个短的面部追踪时间,以防在确定用户移动后有很多动作,并设置一个长的脸部跟踪时间,以防万一有几个动作。 本发明旨在提供一种用于实时检测面部的方法,其能够通过组合面部检测单元和脸部跟踪器来改善面部检测单元的面部检测功能并且有效地调度多线程。
-
公开(公告)号:KR1020160060822A
公开(公告)日:2016-05-31
申请号:KR1020140162399
申请日:2014-11-20
Applicant: 한국전자통신연구원
CPC classification number: G06F17/2735 , G06F17/271 , G06F17/2785 , G06F17/3061
Abstract: 본발명은용언의문형정보를이용한선택제약사전구축방법을제공한다. 상기용언의문형정보를이용한선택제약사전구축방법은용언의문형정보를획득하는단계, 상기획득한용언의문형정보에서용언을추출하고, 상기추출된용언이포함된용례문장을검색하는단계, 상기검색된용례문장의의미를분석하여상기용언과다른의미의용언이포함된용례문장을필터링하는단계, 상기필터링된용례문장의구문을분석하여상기용언과관련없는구성성분을필터링하는단계, 상기필터링된용례문장에서선택제약대상을선택하고, 어휘의미망을이용하여상기선택된선택제약대상의상위경로를추출하는단계및 상기추출된상위경로들의비교를통해상기용언에대한선택제약경로를설정하고, 상기설정된선택제약경로를이용하여상기용언에대한선택제약사전을구축하는단계를포함한다.
Abstract translation: 本发明提供一种使用谓词的句子模式构建选择限制词典的方法,其能够通过使用与句子谓词相关的信息来检测句子的结构和含义。 使用谓词的句型构建选择限制词典的方法包括以下步骤:获取谓词的句型; 从谓词的获取句型中提取谓词以搜索包括提取的谓词的句型; 分析句子模式的含义以过滤包括具有与谓词不同的含义的谓词的句子模式; 分析过滤句子模式的结构,以过滤与谓词无关的组件; 从所述过滤的句子中选择所述选择限制的目标,以通过使用词汇学的网络来提取所述选择限制的所述选定目标的高级路径; 并且通过将所提取的高级路径彼此进行比较来设置谓词的选择限制路径,以及为谓词构建选择限制字典。
-
公开(公告)号:KR101274111B1
公开(公告)日:2013-06-13
申请号:KR1020080131606
申请日:2008-12-22
Applicant: 한국전자통신연구원
IPC: G06Q50/22
CPC classification number: G06Q50/24 , G06F19/00 , G06F19/3418 , G06Q50/22 , G16H10/60
Abstract: 본 발명은 유비쿼터스 환경에서 개인 헬스 디바이스 기반 유니버셜 헬스 플랫폼을 이용한 헬스케어 시스템 및 방법에 관한 것으로, 더욱 상세하게는 사용자가 이용하는 개인 헬스 디바이스에 의해 건강 정보를 측정하고, 측정된 건강 데이터를 자동으로 플랫폼을 거쳐 다양한 헬스케어 서비스 제공에 이용되도록 하기 위해, 적어도 하나 이상의 개인 헬스 디바이스로부터 PHD 표준 개인별 건강 데이터를 수집하여 CDA 표준 데이터로 변환하여 정보 수집 플랫폼으로 전송하는 측정 플랫폼 및 전송된 CDA 표준 데이터를 이용해 복합 CDA 표준 데이터를 생성하여 서비스에 제공하는 정보 수집 플랫폼을 포함한다.
헬스 케어, PHD, HL7 CDA, 플랫폼-
公开(公告)号:KR1020110069347A
公开(公告)日:2011-06-23
申请号:KR1020090126050
申请日:2009-12-17
Applicant: 한국전자통신연구원
CPC classification number: F16F15/02 , B60N2/501 , B60N2/54 , F16F15/002
Abstract: PURPOSE: A vibration absorption system is provided to absorb micro-vibration by offsetting internal and external vibration according to offsetting frequency. CONSTITUTION: A vibration absorption system comprises a vibration sensor(10), a control device(20), and a vibration generating device(30). The vibration sensor is attached to the object material to be sensed. The vibration sensor senses the micro-vibration of the object material. A controller receives the output signal of the vibration sensor and eliminates noise. The controller generates effective offsetting frequency about the valid frequency. The vibration generating device generates vibration according to the offsetting frequency of the control device.
Abstract translation: 目的:提供振动吸收系统,通过根据偏移频率抵消内部和外部振动来吸收微振动。 构成:振动吸收系统包括振动传感器(10),控制装置(20)和振动产生装置(30)。 振动传感器附着到待检测的物体上。 振动传感器检测物体的微振动。 控制器接收振动传感器的输出信号并消除噪音。 控制器生成关于有效频率的有效偏移频率。 振动产生装置根据控制装置的偏移频率产生振动。
-
6.
公开(公告)号:KR1020110034818A
公开(公告)日:2011-04-06
申请号:KR1020090092263
申请日:2009-09-29
Applicant: 한국전자통신연구원
Abstract: PURPOSE: A universal adapter for personal health device standardization of a non-standardized healthcare device and an operating method thereof are provided to easily develop a healthcare device, which enables PHD standard compatibility, at a low cost through a non-standard healthcare device. CONSTITUTION: An adapter interface device(20) communicates with a non-standard healthcare device(10) to collect the measurement data of the non-standard healthcare device. A universal adapter device(30) receives the measurement data from the adapter interface device, and generates a PHD(Personal Health Device) standard message including the received measurement data. According to the PHD standard communication method, the universal adapter device transfers the standard message to a PHD gateway(40).
Abstract translation: 目的:提供用于非标准化医疗保健装置的个人健康装置标准化的通用适配器及其操作方法,以通过非标准保健装置以低成本轻松开发能够实现PHD标准兼容性的保健装置。 构成:适配器接口装置(20)与非标准保健装置(10)通信以收集非标准保健装置的测量数据。 通用适配器装置(30)从适配器接口装置接收测量数据,并且生成包括所接收的测量数据的PHD(个人健康装置)标准消息。 根据PHD标准通信方法,通用适配器设备将标准消息传送到PHD网关(40)。
-
公开(公告)号:KR1020080030138A
公开(公告)日:2008-04-04
申请号:KR1020060095817
申请日:2006-09-29
Applicant: 한국전자통신연구원
CPC classification number: G06F17/30616 , G06F17/30734
Abstract: A method and an apparatus of protein name normalization using ontology mapping are provided to recognize accurately the protein written on the literatures by mapping the protein name recognized in the biological literatures into a normalized protein ontology. A literature recognition part(110) extracts a protein name and species data by accepting biological literatures input. An abbreviated word dictionary DB(130) is composed of pairs of an abbreviated protein names and an original protein names. An abbreviated protein name restoration part(120) restores the abbreviated protein name into the original protein name. A synonym dictionary DB(150) is constructed through the ontology. An inverted index structure DB(160) of the synonym dictionary has an inverted index structure of the synonym dictionary. A protein code analysis part(140) analyzes the protein code by calculating similarity of the protein code by comparing the extracted protein name and the inverted index structure DB of the synonym dictionary. An ontology ID allocating part(190) allocates final ontology IDs by protein name based on the protein code and kind.
Abstract translation: 提供使用本体映射的蛋白质名称归一化的方法和装置,通过将生物学文献中识别的蛋白质名称映射成归一化的蛋白质本体来准确地识别文献中写的蛋白质。 文献识别部分(110)通过接受输入的生物文献来提取蛋白质名称和物种数据。 缩写词典DB(130)由缩写蛋白质名称和原始蛋白质名称的对组成。 缩写蛋白质名称恢复部分(120)将缩写蛋白质名称恢复为原始蛋白质名称。 同义词字典DB(150)是通过本体构建的。 同义字典的反向索引结构DB(160)具有同义字典的反向索引结构。 蛋白质代码分析部分(140)通过比较提取的蛋白质名称和同义词典的反向索引结构DB来计算蛋白质代码的相似性来分析蛋白质代码。 本体ID分配部分(190)基于蛋白质代码和种类,通过蛋白质名称分配最终本体ID。
-
公开(公告)号:KR1020070060993A
公开(公告)日:2007-06-13
申请号:KR1020060024786
申请日:2006-03-17
Applicant: 한국전자통신연구원
Abstract: A method and a system for verifying protein-protein interaction with text mining are provided to avoid a duplicated experiment by utilizing knowledge proved through documents before the estimated protein-protein interaction is experimentally proved and generate a measure for estimating performance of an estimation system by verifying a result of the estimation system. An ontology database(160) stores protein-protein interaction and hierarchical structure information among proteins. A text mining part(120) extracts the protein-protein interaction from protein-related documents by using a text mining method. An ontology mapper(130) maps the extracted protein-protein interaction to an ontology ID by using the ontology database. An information filter(140) filters the information having high weight based on a frequency of the information and an effect factor of the protein-related documents among the mapped protein-protein interaction information. An information indexer indexes protein-related document/sentence, ontology ID, protein-protein interaction, and precision information.
Abstract translation: 提供了一种用于验证与文本挖掘的蛋白质 - 蛋白质相互作用的方法和系统,以在通过实验证明估计的蛋白质 - 蛋白质相互作用之前利用通过文献证明的知识来避免重复的实验,并通过验证来生成估计估计系统的性能的度量 是估算系统的结果。 本体数据库(160)在蛋白质之间存储蛋白质 - 蛋白质相互作用和层次结构信息。 文本挖掘部分(120)通过使用文本挖掘方法从蛋白质相关文档中提取蛋白质 - 蛋白质相互作用。 本体映射器(130)通过使用本体数据库将提取的蛋白质 - 蛋白质相互作用映射到本体ID。 信息过滤器(140)基于所述信息的频率和所映射的蛋白质 - 蛋白质相互作用信息之间的蛋白质相关文档的影响因子来过滤具有高权重的信息。 信息索引器索引蛋白质相关文档/句子,本体ID,蛋白质 - 蛋白质相互作用和精确度信息。
-
-
-
-
-
-
-
-