Photobook: content-based manipulation of image databases
International Journal of Computer Vision
VisualSEEk: a fully automated content-based image query system
MULTIMEDIA '96 Proceedings of the fourth ACM international conference on Multimedia
Neural Networks: A Comprehensive Foundation (3rd Edition)
Neural Networks: A Comprehensive Foundation (3rd Edition)
PicToSeek: combining color and shape invariant features for image retrieval
IEEE Transactions on Image Processing
A detailed survey on query by image content techniques
ICNVS'10 Proceedings of the 12th international conference on Networking, VLSI and signal processing
Content-based medical image retrieval in peer-to-peer systems
Proceedings of the 1st ACM International Health Informatics Symposium
Hi-index | 0.00 |
The overall objective of this paper is to present an extended application of Content-Based Image Retrieval (CBIR) over distributed (decentralized) image databases. Traditional image retrieval system design has implicitly relied on a local (centralized) query server, such as IBM's QBIC [1], Columbia's VisualSEEk [2], MIT's PhotoBook [3], and UCSD's Viagem™ [4]. With the growing popularity of the internet, however, the focus of the research in this area has been shifted toward content query over distributed databases. Ng et al. [5] has studied a peer-clustering model for the query with the assumption that the image collection at each peer node falls under one category. Even though, this assumption is effective for preliminary studies, it is unable to implant the practical end-user behaviors. Lee et al. [6] has introduced a novel approach to study practical scenarios where multiple image categories exist in each individual database in the distributed storage network. This approach is proven to be an effective method to improve retrieval precision via identifying the community neighborhood who shares similar content collection. The main focus of this paper is to study behavior of a CBIR engine in an interactive distributed environment. In the proposed approach, the query image is sent to all registered databases in the network. Response of each database is then collected and transferred to a local server where a supervised relevance identification approach is applied to identify final outcome of the search. Response of each database is quantified via estimating the statistical resemblance of top image candidates to the existing query image. Comprehensive experiments demonstrate feasibility of the proposed methodology.