基于多重特征关联的视频GIS数据多模态检索方法

VideoGIS is the new generation multi-media GIS that fuses GIS and geographic videos, it is extensively used in public safety, transportation management, public service and other applications. With the rapid development of smart cities and the advance of city safety and defense demands, the question that how to accurately discover and retrieve the data that users request faces a series of bottle-neck problems. Traditional GIS merely treats geographic videos as multi-media attributes, thus it is difficult to perform deep association analysis on multi-modal VideoGIS data under geographic context constraints. Aimed at this problem, this research studies video GIS data associative retrieval methods based on multi features association model. The main contents contain: (1) the representation and construction of VideoGIS data multi features association model; (2) the VideoGIS data associative retrieval methods based on multi features association model and user conscious feedback; (3) the cost model of multi-modal VideoGIS data retrieval. This study can provide powerful support for the integrated association analysis of VideoGIS data that fuses semantic features, visual features and spatio-temporal features under geographic context constraints and enriches the theories and methods of VideoGIS; the study expands the scope and depth of VideoGIS applications and develops and increments the potential value of ‘VideoGIS Big Data’.

视频GIS是GIS与地理视频相融合产生的新一代多媒体GIS，在公共安全、交通管理以及大众服务等方面中得到广泛应用。随着智慧城市建设的发展和城市安防要求的提高，如何从视频GIS大数据中准确发现和获取用户所需的数据面临一系列瓶颈问题。针对传统GIS仅把地理视频当作多媒体属性，难以在地理环境约束下对多模态视频GIS数据进行深度关联分析的现状，本项目研究基于多重特征关联模型的视频GIS数据关联检索方法，主要内容包括：（1）视频GIS数据多重特征关联模型的表达与构建；（2）基于多重特征关联模型和用户感知反馈的视频GIS数据多模态关联检索方法；（3）视频GIS数据多模态检索代价模型。本项目的研究为地理环境约束下融合语义特征、视觉特征和时空特征的视频GIS数据一体化关联分析提供了有力支撑，丰富了视频GIS的理论和方法，拓展了视频GIS应用的广度和深度，使视频GIS大数据的潜在价值得以充分发挥和增值。

视频GIS是GIS与地理视频相融合产生的新一代多媒体GIS，在公共安全、交通管理以及大众服务等方面中得到广泛应用。随着智慧城市建设的发展和城市安防要求的提高，如何从视频GIS大数据中准确发现和获取用户所需的数据面临一系列瓶颈问题。本项目提出了融合语义特征、视觉特征和时空特征的视频GIS多特征关联数据模型，研究了视频GIS数据多模态检索方法，构建了视频GIS数据检索性能代价模型，并探索了平面视频空间化的方法以支撑视频GIS的耦合分析与检索结果的友好显示。本项目的研究丰富了视频GIS的理论和方法，拓展了视频GIS应用的广度和深度，使视频GIS大数据的潜在价值得以充分发挥和增值。