公开(公告)号：KR100738423B1

公开(公告)日：2007-07-12

申请号：KR1020060016029

申请日：2006-02-20

Applicant: 삼성전자주식회사

Inventor： 정진국 ,  김상균 ,  정용주 ,  황의현 ,  황두선 ,  김지연 ,  고산 ,  정철곤 ,  이용

IPC: H04W4/06 ,  H04W84/10 ,  H04W4/00

Abstract: A method and a system for providing a broadcast service using a home server and a mobile terminal are provided to allow a user to watch only a desired scene in a recorded broadcast program. A broadcast receiving unit(110) receives broadcast data. A summary generating unit(124) analyzes the received broadcast data and generates a broadcast summary data according to a request for recording a broadcast. A database(127) stores the generated broadcast summary data and the received broadcast data. A home server input/output management unit(129) receives the request for recording the broadcast or a request for selecting a summary from a mobile terminal of a user, selects the stored broadcast summary data, and manages so as to output a streaming service for the selected broadcast summary data to the mobile terminal of the user.

Abstract translation: 提供了一种使用家庭服务器和移动终端提供广播服务的方法和系统，以允许用户只在所记录的广播节目中观看期望的场景。 广播接收单元（110）接收广播数据。 摘要生成单元（124）根据记录广播的请求分析接收到的广播数据并生成广播摘要数据。 数据库（127）存储所生成的广播摘要数据和接收到的广播数据。 家庭服务器输入输出管理部（129）从用户的移动终端接收记录广播的请求或选择摘要的请求，选择存储的广播摘要数据并进行管理，输出流媒体业务 将选择的广播摘要数据发送给用户的移动终端。

公开(公告)号：KR100804678B1

公开(公告)日：2008-02-20

申请号：KR1020070000957

申请日：2007-01-04

Applicant: 삼성전자주식회사

Inventor： 정진국 ,  김지연 ,  김상균 ,  고산

IPC: H04N21/845 ,  G06T7/20

CPC classification number: G06K9/00295 ,  G06F17/30793 ,  G06K9/00718 ,  G06K9/00765

Abstract: A method and a system for classifying scenes according to video persons are provided to classify persons in a scene unit by using temporal information in video data, thereby providing a story view by person. A method for classifying scenes according to video persons comprises the following steps of: detecting a face for inputted video frames(410); detecting shot conversion for the inputted video frames(420); extracting a person representative frame within a shot(430); performing person clustering based on temporal information in the extracted person representative frame(440); detecting scene conversion by separately comparing a person and a background based on face extraction information(450); and performing scene clustering by person by merging similar clusters from the extracted person representative frame(460).

Abstract translation: 提供根据视频人员对场景进行分类的方法和系统，通过使用视频数据中的时间信息对场景单元中的人进行分类，从而提供人的故事视图。 根据视频人员对场景进行分类的方法包括以下步骤：检测输入的视频帧的面部（410）; 检测所输入的视频帧的镜头转换（420）; 提取镜头内的人物代表帧（430）; 基于提取的人代表帧中的时间信息执行人物聚类; 通过基于面部提取信息单独比较人与背景来检测场景转换（450）; 以及通过从提取的人代表帧合并相似的聚类来执行人的场景聚类（460）。