项目名称: 监控视频数据中知识发现方法研究
项目编号: No.61472166
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 其他
项目作者: 叶飞跃
作者单位: 江苏理工学院
项目金额: 81万元
中文摘要: 监控视频数据中的知识发现对于公共安全、交通管理等方面具有重要意义,然而,监控视频具有数据量大、价值密度低的大数据特征,加上多点监控视频数据具有时空属性,因此,监控视频数据知识发现中的一些关键问题亟待解决。本课题将在监控视频数据中的时空频繁模式、时空关联规则、近似时空依赖关系发现方法上有所创新,并形成监控视频信息提取与数据挖掘算法作为有机整体考虑的研究特色。研究内容主要包括:1)近似时空依赖挖掘理论和模型、算法研究;2)时空频繁模式、时空关联规则挖掘理论、模型、算法研究;3)监控视频数据中高效的人脸检测方法比较和选择研究;4)快速和高精确率的人脸验证方法比较及改进研究;5)一种人脸特征数据编码分类方法研究;6)具有时空属性事件存储模型和有关事件存储方法、存储结构的研究;7)面向人员出现事件的时空频繁模式和时空关联规则挖掘原型系统研发。
中文关键词: 知识发现;数据挖掘;监控视频;大数据
英文摘要: The knowledge discovery to surveillance video data is important to many areas of our society like security and traffic management. However surveillance video data is big data with large data volume and low value density feature and with spatio-temporal attribute.Therefore many problems relating to the knowledge discovery from the type of data still need solutions or better solutions. This project aims to discovery methods innovation for discovering frequent events, association rules and approximate dependencies relating to temporal and spatial features from video data and the research characteristics will be formed surveillance video information extraction and data mining algorithms considered as an organic whole. The problems to be investigated by the project include 1) the development of theories and models and algorithms of approximate temporal and spatial dependencies to surveillance video data; 2) the development of theories and models of temporal and spatial frequent patterns, associations rules to surveillance video data;3) research on comparison and selection method for face detection in video data efficiently;4)research on comparison and improvement method for face verification in video data efficiently and high accurate rate of face;5)classification method for face feature data coding; 6) the design of the methods and structures for storing and retrieving temporal and spatial events; 7) the development of a prototype for discovering people's occurrence temporal and apatial frequent patterns and temporal and spatial association rules.
英文关键词: Knowledge Discovery;Data mining;Surveillance video;Big data