Browsing by Author "Cheng, PJ"

Now showing 1 - 2 of 2

A new content-based access method for video databases
(1999-09-01)
Cheng, PJ
;
Wang, WP
;
資訊工程學系
;
Department of Computer Science
This paper presents a novel video data model and a nested annotation language for describing complex information of video data. In contrast to conventional approaches, the proposed model classifies different video materials of interest to users into different representation frameworks according to their individual properties. It makes the model flexible and capable of sharing video materials. The nested annotation language effectively describes scenarios in video data and can be effectively analysed. With the assistance of domain knowledge and index organizations this investigaton also develops algorithms to effectively process five types of familiar video queries: semantic query, temporal quaery, similar query, fuzzy query and hybrid query. In addition, a SQL-like query language for video content retrival is provided. Experimental results indicate that combining the concepts of Bayesian networks and inheritance of attributes by context significantly improves the content based retrieval of video data. Moreover a prototype system based on proposed model has been implemented. (C) 1999 Published by Elsevier Science Inc. All rights reserved.
Composition and retrieval of visual information for video databases
(2001-12-01)
Cheng, PJ
;
Yang, WP
;
資訊工程學系
;
Department of Computer Science
This paper presents a new visual aggregation model for representing visual information about moving objects in video data. Based on available automatic scene segmentation and object tracking algorithms, the proposed model provides eight operations to calculate object motions at various levels of semantic granularity. It represents trajectory, color and dimensions of a single moving object and the directional and topological relations among multiple objects over a time interval. Each representation of a motion can be normalized to improve computational cost and storage utilization. To facilitate query processing, there are two optimal approximate matching algorithms designed to match time-series visual features of moving objects. Experimental results indicate that the proposed algorithms outperform the conventional subsequence matching methods substantially in the similarity between the two trajectories. Finally, the visual aggregation model is integrated into a relational database system and a prototype content-based video retrieval system has been implemented as well. (C) 2001 Academic Press.