From oscar Mon Jan 29 11:44:26 1996 Flags: 000000000001 Received: (from oscar@localhost) by DB.Stanford.EDU (8.7.1/8.7.1) id LAA28288 for gio@db.stanford.edu; Mon, 29 Jan 1996 11:44:25 -0800 Date: Mon, 29 Jan 96 11:44:24 PST From: Oscar Firschein To: gio@DB.Stanford.EDU Subject: Revised Refs (last one for a long time) Message-ID: CS545I References

CS 545I - Advanced Image Databases, W 95/96

References

1a - Web sites re Query by Image Content QBIC System http://wwwqbic.almaden.ibm.com/~qbic/qbic.html Try out the IBM QBIC system. Comments for improvement are welcome. Berkeley Digital Library Project, http://elib.cs.berkeley.edu/cypress/ Try out the Cypress image retrieval system. Not a very friendly front end. ImageEngine http://www.cml.upmc.edu/ImageEngine.html A medical image retrieval system being developed by the Section of Medical Informatics, Dept of Medicine, U of Pittburgh. MIT Photobook http://www.media.mit.edu/groups/casr/picard.html Also visit sites of Profs. Pentland and Lippman. Jacob (Just A COntent-Based) Query system for Video DB , U of Palermo http://wwwcsai.diepa.unipa.it:80/research/projects/jacob Describes system and lets you try some queries. Not a very friendly front end. Virage Corp. http://www.virage.com Lets you try queries. Virage is a small image retrieval company in the San Diego area. At the present time, the only other commercial offering besides IBM QBIC. 1b - Related to images (not content-based, but interesting) Yahoo Search engine http://www.yahoo.com Search for multimedia, pictures,archives; online image archives OR picture archives. Getty Art History Information Program, http:www.ahip.getty.edu/ahip How to build image databases; image capture; resources for indexing art. 1c - University work or courses MIT Media Lab has many good projects, especially from Profs. Picard, Pentland, and Lippman Prof. Sclaroff, Boston university http://cs-pub.bu.edu/faculty/sclaroff/courses/cs835-95/projects.html Description of student projects Sean Landis Term Paper http://www.tc.cornell.edu/~landis/ A very informative student term paper describing a system for "Content-based image retrieval systems for interior design." 1d - Database stuff IBM DB2 database http://www.software.ibm.com Mouse "Data Managemen" icon and then DB2 under that. Example of database offering on the Web that integrates SQL and WWW, both at the user interface and internally. Getty Museum http://simpr1.compapp.dcu.ie/images/query.html Natural language interface to image database (not query by image content) 1e- Just for fun Browsing for art: http://www.cmp.ucr.edu photography exibits http://www.metmuseum.org Metropolitan Museum of Ar http://www.rosprint.ru/art/museum/pushkin Pushkin Museum http://sunsite.unc.edu/wm WebMuseum, Paris http://www.echonyc.com/~whitney Whitney Museum of Art 2- Books The book we recommend has several sections on content based retrieval for image and video DB - the only book to address this topic
Furht, Smoliar, and Zhang: Video and Image Processing in Management Systems; Florida Atlantic University, Boca Raton, FL. Kluwer Academic Publishers, 1995 (ABOUT $150!)

3-Collections and Journals Storage and Retrieval for Images and Video Databases II; IS&T/SPIE Symposium on Electronic Imaging Science and Technology,Feb 95, SPIE Vol. 2185. Storage and Retrieval for Images and Video Databases III; IS&T/SPIE Symposium on Electronic Imaging Science and Technology,Feb 95, SPIE Vol. 2420. Multimedia Systems Journal Proc. ACM Multimedia '93, Anaheim, CA, Aug.1993. Proc. ACM Multimedia '94, San Francisco CA, Oct.1994. 4 - Papers primarily on still imagery:

Pentland et al: Photobook: Tools for Content-Based Manipulation of Image Databases; Perceptual Computing Section, The Media laboratory, MIT, Cambridge, MA 02139, SPIE, Storage and Retrieval Image and Video Databases II, No. 2185, Feb 1994 Describes Photobook, a set of interactive tools for browsing and searching images and image sequences. Direct search on image content is made possible by the use of semantics-preserving image compression. The paper describes three Photobook tools, one that allows search based on grey-level appearance, one that uses 2-D shape, and a third that allows search based on textural properties.
Petkovic et al: Recent Applications of IBM's Query by Image Content (QBIC); IBM Almaden Research Center, San Jose, CA 95120-6099, SAC '96, 5 pp. Present a brief overview of QBIC technology and products, and a summary of some applications.
Faloutsos et qal: Efficient and Effective Querying by Image Content; Journal of Intelligent Information Systems, 3, 231-262(1994) An excellent technical paper describing the IBM QBIC system, giving details of the image features and distance functions, indexing details, and experiments on the effectiveness of QBIC. Two problems are specifically addressed, non-Euclidean distance measures and the high dimensionality of feature vectors.
no author listed: IBM Database 2 Version 2, Relational Extenders; 4pp., 1995? A brief description of several extenders forIBM's DB2, including text, image, video, audio, and fingerprint extenders.
Okon, C.: IBM's Image Recognition Tech for Databases at Work: QBIC or not QBIC?; Advanced Imaging Magazine, pp. 63-65, May, 1995. A short, breezy, layman's view of QBIC.
Holt, B. & Hardwick, L.: Retrieving Art Images by Image Content: the UC Davis QBIC Project; Aslib Proceedings, Vol 46, n. 10, October, 1994, pp. 243-248. Describes the construction of a pilot database for retrieving art images based on what they look like, rather than relying on text indexing. Uses IBM QBIC as the search engine. Preliminary findings of the project are presented.
Hafner et al: Efficient Color Histogram Indexing for Quadratic From Distance Functions; IBM Almaden Research Center, San Jose, Ca 95120-6099, IEEE Trans. PAMI, Vol. 17, N.7, pp. 729-736, July 1995. In image retrieval based on color, the weighted distance between color histograms of two images, represented as a quadratic form, may be defined as a match measure. However, this distance measure is computationally expensive, and it operates on high-dimensional features. This paper proposes the use of low-dimensional, simple-to-computer distance measures between the color distributions, and shows that these are lower bounds on the histogram distance measure .
Aigrain, P.: Image and Sound Digital Libraries Need More Than Storage and Networked Access; Institut de Recherche en Informatique de Toulouse, Universite Paul Sabatier, 118 route de Narbonne, 31062 Toulouse Cedex, France, 7 pp. Good general paper that argues for the need for content-based retrieval, content-skimming and abstracted representations, and content-based user interfaces for interactive viewing and listening.
Pentland et al: View-Based and Modular Eigenspaces for Face Recognition; Perceptual Computing Section, The Media laboratory, MIT, Cambridge, MA 02139 IEEE CVPR, 1994, 11pp. This paper describes experiments with eigenfaces for recognition, verification, and interactive search in a large-scale face database. In mathematical terms, they find the principal components of the distribution of faces, or the eigenvectors of the covariance matrix of the set of face images. These eigenvectors can be thought of as a set of features which together characterize the variation between face images. The eigenvector can be displayed as a sort of ghostly face which is called an eigenface. Questions of overall accuracy and sensitivity to orientation, scale, and illumination are addressed.
Tamura et al: Image Database Systems: a Survey; Electrotechnical Laboratory, 1 1-4 Umezono, Sakura-mura, Niihar-gu, Ibaraki 305, Japan Pattern Recognition, vol. 17, n. 1, 1984, pp 29-43. Although an old paper, this survey still provides an excellent summary of image database systems.
Jacobs et al: Fast Multiresolution Image Querying; Department of Computer Science and Engineering University of Washington, Seattle, WA 98195-2350 ACM SIGGRAPH 1995, 11 pp., Computer Graphics Proceedings, Annual Conference Series, 1995. Presents a method for searching in an image database using a query image that is similar to the intended target. The query image may be a hand-drawn sketch or a scan of the image to be retrieved. The searching algorithm makes use of multiresolution wavelet decompositions of the query and database images. The coefficients of these decompositions are distilled into small "signatures" for each image, and an "image query metric" operates on these signatures. Experiment with hundreds of queries in databases of 1000 and 20,000 images show dramatic improvement, in both speed and success rate, over using conventional norms.
Liu et al: Periodicity, Directionality and Randomness: Wold Features for Perceptual Pattern Recognition; Perceptual Computing Section, The Media laboratory, MIT, Cambridge, MA 02139 Proceedings of the 12th ICPR 6pp., Jerusalem, Israel, October, 1994. In applications such as image retrieval, it is important that features used by the system in pattern comparison provide good measures of "perceptual similarity." A new set of features and an image model are presented based on the three mutually orthogonal components produced by the 2-D Wold decomposition of random fields. The effectiveness of the new Wold features for retrieving perceptually similar natural textures is demonstrated.
Cawkell, A.: Picture-queries and Picture Databases; Citech, Ltd., Iver, Bucks, UK, Journal of Information Science, 19 (1993) pp. 409-423, pub. Elsevier. A non-mathematical exposition of existing image database systems, indexing by query pictures, examples of query-image systems, and database organization.
Gong et al: An Image Database System with Content Capturing and Fast Image Indexing Abilities; School of Electrical and Electronic Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 2263. IEEE Int. Conf on Multimedia, Computing, and Systems, Boston, May 1994, 10 pp. Describes a system very similar to QBIC, using both keywords and image properties in the query.
Barber et al: Efficient and Effective Querying by Image Content; IBM Almaden Research Center, San Jose, Ca 95120-6099, Journal of Intelligent Information Systems, 3, pp 231-262 Kluwer, 1994. A set of features and similarity measures allowing query by image content are described, along with the QBIC system. The effectiveness of the system with normalized precision and recall experiments is demonstrated. The problems of non-Euclidean distance measures and high dimensionality of feature vectors are specifically addressed
Jain, R. et al: Workshop Report: NSF-ARPA Workshop on Visual Information Management Systems, 20 pp.; Electrical and Computer Engineering, University of California at San Diego, La Jolla, CA 92093-0407. Summarizes results of a workshop held on June 19, 1995 in the areas of visual information systems, image databases, video databases, and infrastructure. Issues and research topics for each of these areas are indicated. For bibliography, see http://www.virage.com/vir-res/

Hermes et al Image retrieval for information Dept of Comp. Sci., U of Bremen, Germany Storage and Retrieval for Images and Video Databases III; IS&T/SPIE Symposium on Electronic Imaging Science and Technology,Feb 95, Vol. 2420, pp.281-290. Project IRIS (Image Retrieval for Information Systems) combines methods and technique in computer vision and AI to automatically generate content descriptions of images in a textual form. IRIS then retrieves the images by means of text retrieval using the IBM SearchManager/6000. 3 - Papers primarily on video retrieval:
Zhang et al: Automatic parsing of News Video; Institute of Systems Science, National University of Singapore, IEEE Conference on Multimedia Computing and Systems, Boston, May, 1994. Video content parsing is possible when one has an a priori model of the structure of a video based on domain knowledge. This paper presents work on using domain knowledge to parse the content of news video programs. Approaches have been developed for locating and identifying frame structure models based on temporal and spatial structure of news video data, along with algorithms to apply these models in parsing news video. Experimental results are discussed in detail.
Ahanger et al Video query formulation Dept of Elect, Comp, and Systems Eng., Boston Univ. and Siemans, Princeton, NJ Storage and Retrieval for Images and Video Databases III; IS&T/SPIE Symposium on Electronic Imaging Science and Technology,Feb 95, Vol. 2420, pp.281-290. Reviews current video retrieval, describes video attributes and the issues related to video data. Identifies queryable attributes unique to video data, namely audio, temporal structure, motion, and events. The approach is based on visual query methods to describe predicates interactively while providing feedback that is as similar as possible to the video data. An initial prototype of the visual query system for video data is presented.
Rowe et al Indexes for user access to large video databases Computer Science EECS, U of Calif, Berkeley, CA Storage and Retrieval for Images and Video Databases II; IS&T/SPIE Symposium on Electronic Imaging Science and Technology, Feb 95, Vol. 2185, pp.150-161. This paper describes the design and implementation of a metadata database and query interface for locating a video of interest in a large video database. The design was guided by the results of interviews with users. A mixed-mode query interface was built that allow a user to select a set of videos and/or frame sequences and incrmentally modify the answer set.