| 摘要 |
The invention relates to a method and process dealing with the development of informatics applications supporting the efficient indexing, organizing and smart visual content-based search of images and video image-series by using the following techniques: By unsupervised sectioning of compressed moving and unmoving images in regions corresponding to time-space objects with combinational use of colour, position, movement, and texture of pixels; by extraction, matching and similarity of the searchfeatures from time-space objects; by semantic analysis allowing the translation of low-level to high-level search features. The matching of low-level and high-level features is established either automatically by statistic treatment or by integration of same features to a knowledge base describing the search field (e.g. ontology) with high-level meanings. In this way, the automatic matching of the visual features of images and video image-series with higher-level meanings is achievable by search carried out by use of text and meanings stored in a knowledge database. The informatics applications referring to the efficient organizing and smart search of images and video image-series involve: organizing and visual materials search in enterprises such as radio and television stations, digital subscription television, advertising and production companies, graphic arts, photographers, informatics enterprises developing related services, Internet portals, newspapers websites, press agencies,websites of large-sized enterprises, libraries, medical and military applications, search through web, personal digital images and video image-series etc. |
