Chang, S.-F. and J. R. Smith, 1997. Visual Information Retrieval from Large Distributed Online Repositories. Communications of the ACM, 40(12): 63-71.  Home/Search   Document Details and Download   Summary   ACM   TOC   Related Articles   Check

....of images that are available. As one example, AltaVista has over 29 million images in its index as of May 1, 2002. 1] Another important issue is finding ways of searching through the enormous pool of information when users may be unable to accurately describe exactly what they are looking for [5]. A recent survey paper on Internet search identifies three important activities that are necessary for successful search [11] The first is indexing, which in the case of images, includes finding appropriate features to describe an image, as well as cataloging those features and the image s ....

....the authors determined that text based keyword searching for images increased the cognitive load on the user, requiring more work to visually inspect the results, determine relevance, and decide whether and how to reformulate the query. Specialized multimedia retrieval systems such as WebSEEk [5, 14] and AMORE [12] combine both textbased and content based retrieval of images. WebSEEk, for example, allows users to first narrow down their searches by selecting a category from a semi automatically defined hierarchy. Images are pre assigned to categories based on textual cues such as file names ....

[Article contains additional citation context not shown here]

Chang, Shi-Fu, John R. Smith, Mandis Beigi, and Ana Benitez, "Visual Information Retrieval from Large Distributed Online Repositories," Communications of the ACM, vol. 40, December 1997, pp. 63-71.

....to identify the desired video manually. The title and description de nitely give the user a general idea about the video. However, a preview is usually required to give a better picture of the entire video. Conventional pipelining is unable to support this feature. Many digital video libraries [1, 2, 3, 4] have to use an extra preview le for each video. The preview le is created in advance by extracting only important frames from the original video using some image processing techniques. This approach has two drawbacks. First, it requires more storage space on the servers. Second, downloading the ....

S.-F. Chang, J. R. Smith, M. Beigi, and A. Benitez. Visual information retrieval from large distributed online repositories. Commun. ACM, 40(12):63-71, Dec. 1997.

....of video clips that is amenable to browsing. In particular [19] used graphs to represent the temporal relationships between scenes. The camera motion motion and gross statistics of motion in a given shot have also been used as descriptions of shots [8] Object based representations such as [2], in which one applies some sort of segmentation and or tracking to find potentially moving objects, are ways of more fully utilizing the spatialtemporal nature of video. 2] 4] and[13] all detected and indexed the trajectories of objects within shots. These approaches are low level and local in ....

....statistics of motion in a given shot have also been used as descriptions of shots [8] Object based representations such as [2] in which one applies some sort of segmentation and or tracking to find potentially moving objects, are ways of more fully utilizing the spatialtemporal nature of video. [2], 4] and[13] all detected and indexed the trajectories of objects within shots. These approaches are low level and local in nature, however, and only provide indexing capabilities for rather precise queries. Also, they provide access to video clips of a very small time scale. A few recent methods ....

S.-F. Chang et al. Visual information retrieval from large distributed online repositories. Comm. ACM, 1(1), 1996.

....domain, person culture, expert level, etc. For example, a query on heart may retrieve all heart X rays in medical domain and a photo of two lovers in pictorial one. Because manual procedures are used (key words with QBIC [3] or legend (text) indexed classically ( Mec 95] Leu 95] Sut 97] Cha 97] Yan 97] Li 98] and [Jia 99] the golden age of this approach was brief. The appropriate approach in medical domains must take into account firstly the complexity observed when describing semantic content of images (objects, their relations and properties) and secondly the graphical aspect ....

Chang, S. F. et. al., "Visual Information Retrieval from Large Distributed Online Repositories", Comm. ACM, Vol. 40, Dec 1997, pp. 63-71.

....wonder: how do they manage to retrieve exactly the video I want 6. Video databases on the Internet Making a VDBS accessible through the Internet, particularly on the Web, extends its usefulness to users anywhere in the world at the expense of new design constraints which are addressed below [11]: Visual information on the Web is highly distributed, minimally indexed, and schema less. The query and retrieval stages have no control over the cataloguing process and must rely on possible metadata stored in HTML tags associated with the images and video clips. In order to keep the ....

S.-F. Chang, J. R. Smith, M. Beigi, and A. Benitez, "Visual Information Retrieval from Large Distributed Online Repositories". Communications of the ACM, Vol. 40, No. 12, December 1997.

....show that D EM has a satisfactory performance in image retrieval applications. D EM algorithm has the potential to many other applications. 1 Introduction Recent years have witnessed a rapid increase of the volume of digital image collections, which motivates the research of image retrieval [2, 9, 11]. Early research of image retrieval is searching by manually annotating every image in a database. However, these text based techniques are impractical for two reasons: large size of image databases and subjective meanings of images. To avoid manual annotating, an alternative approach is ....

....feature space instead of raw image space. Physical features and mathematical features are two typical representations. Many research e#orts have been made to extract physical features such as color features, texture features, edge features, structure features, or combination of these features [2, 7, 12]. However, images are too rich to represent by these physical features. An alternative representation is mathematical features, which only performs dimension reduction in mathematical senses. Principal component analysis (PCA) is a typical technique to obtain such mathematical features [14] ....

S.Chang, J.Smith, M.Beigi and A.Benitez, "Visual Information Retrieval from Large Distributed Online Repositories", Communications of ACM, Dec. pp.12-20, 1997

....manual annotating large image databases, an alternative approach of retrieving images is content based image retrieval (CBIR) by which images would be indexed by their visual contents such as color, texture, shape, etc. Many research efforts have been made to extract these lowlevel image features [1, 4, 9], evaluate distance metrics [7, 10] and look for efficient searching schemes [11, 14] However, images are too rich to represent by these lowlevel physical features. An alternative representation is mathematical features, which only performs dimension reduction in mathematical senses. Principal ....

S.Chang, J.Smith, M.Beigi and A.Benitez, "Visual Information Retrieval from Large Distributed Online Repositories", Communications of ACM, Dec. pp.1220, 1997

....how much time person spends looking at certain types of objects, etc. but also in determining what is deemed as important during the process (i.e. areas looked at are probably more important than areas not looked at) In the last few years, research in the field of Content Based Retrieval [3] has focused on facilitating access to multimedia information (e.g. images, video, etc. in large digital databases. In particular, there has been a strong interest in being able to automatically classify multimedia data. Images and video, for example, can be placed into categories depending on ....

Chang, S.-F., Smith, J.R., Beigi, M. and Benitez, A. "Visual Information Retrieval from Large Distributed On-line Repositories," Communications of the ACM, 40(12):63-71, December, 1997.

No context found.

S.-F. Chang, J.R. Smith, M. Beigi and A. Benitez, "Visual Information Retrieval from Large Distributed On-line Repositories," Communications of the ACM, 40(12):63-71, December, 1997.

....focus on di erent sections of users and content. As a result, the indexing features and the subject taxonomies are also di erent, causing the concern of interoperability. Several recent e orts in standards have started to address this issue [2, 163] Several research systems on image metaservers [13, 22, 95] have also investigated frameworks for integrated access to distributed image libraries. 5.4. High Dimensional Indexing A by product of the web expansion is the huge collection of images. Most currently existing research prototype systems only handle hundreds or at most a few thousand of images, ....

S.-F. Chang, J. R. Smith, M. Beigi, and A. Benitez. Visual information retrieval from large distributed online repositories. Communications of ACM, Special Issue on Visual Information Retrieval, 40(12):12-20, Dec 1997.

.... Even though some of these measures are difficult to quantify for a human observer, these global low level features have been successfully used in various content based retrieval systems to organize the contents of a database for browsing and to perform query by example (QBIC, VisualSEEk, Virage) [6]. 5.1.4 Local Structure In contrast to Global Structure, which does not provide any information about the individual parts of the image or the video sequence, the Local Structure level is concerned with the extraction and characterization of the components. At the most basic level, those ....

....box) Figure 5c shows images in which attributes of this type may be of importance. In x rays and microscopic images there is often a strong concern for local details. Such elements have also been used in content based retrieval systems, mainly on query by usersketch interfaces such as VisualSEEk [6]. The concern here is not with objects, but rather with the basic elements that represent them and with combinations of such elements a square, for example, is formed by four lines. 5.1.5 Global Composition At this level, we focus on the specific arrangement or composition of the basic elements ....

S.-F. Chang, J. R. Smith, M. Beigi, and A. B. Benitez, "Visual Information Retrieval from Large Distributed On-line Repositories", Communications of ACM, Vol. 40, No. 12, pp. 63-71, Dec 1997. J.

....metasearch engine using the proposed image DS. Metasearch engines act as gateways linking users automatically and transparently to multiple search engines. Our metasearch engine, MetaSEEk [1] explores the issues involved in querying large, distributed, on line visual information systems [2]. In this section, we will describe how the proposed image DS will be used to enhance the metasearch systems. MetaSEEk is designed to intelligently select and interface with multiple on line image search engines by ranking their performance for different classes of user queries. The three main ....

S.-F. Chang, J. R. Smith, M. Beigi, and A. B. Benitez, "Visual Information Retrieval from Large Distributed On-line Repositories", Communications of the ACM, Vol. 40, No. 12, pp. 63-71, Dec. 1997.

....image search systems can be greatly improved. There has been substantial progress in developing powerful tools which allow users to specify image queries by giving examples, drawing sketches, selecting visual features (e.g. color and texture) and arranging spatial temporal structure of features [27]. Much success has been achieved, particularly in specific domains, such as sports, remote sensing, and medical applications. New challenges remain in applying the above content based image search tools to meet real user needs. Our experience indicates that use of the image search systems varies ....

S.-F. Chang, J. R. Smith, M. Beigi, and A. Benitez. Visual information retrieval from large distributed online repositories. Communications of ACM, Special Issue on Visual Information Management, 40(12):63-- 71, Dec. 1997.

....index visual content. Some of these techniques are based on similarity or query by sketch approaches (e.g. an image that looks like another one; an image that resembles a drawing) Examples of systems that use similarity search or query by sketch techniques include QBIC, and VisualSEEk [4]. Other recent work has focused on the automatic extraction of higher level descriptions of the visual content, via classification. In that scenario, the image or video is automatically placed into a semantic category. Examples of this approach include the classification of images according to ....

S.-F. Chang, J.R. Smith, M. Beigi and A. Benitez, "Visual Information Retrieval from Large Distributed On-line Repositories", Communications of the ACM, December, 1997. ACM.

....annotations are used for indexing a cataloguer manually assigns a set of key words or expressions to describe an image. Users can then perform text based queries or browse through manually assigned categories. In contrast to text based approaches, recent techniques in content based retrieval [38], have focused on indexing 3 images based on their visual content. Users can perform queries by example (e.g. images that look like this one) or user sketch (e.g. image that looks like this sketch) More recent efforts attempt automatic classification of images based on their content: a system ....

....[24] 25] classification [26] 29] query analysis [20] 21] and indexing schemes [27] 30] among others. Most work in content based retrieval has focused mainly on using low level features for automatic classification based global features [56] 59] query by example (QBIC, Virage, VisualSEEk) [38], and query by user sketch [42] More recent work has been performed on object based classification [44] There have also been recent efforts related to the organization of multimedia data. Some of that work includes [40] 41] 46] 55] and [48] In addition, the development of the MPEG 7 standard ....

[Article contains additional citation context not shown here]

S.-F. Chang, J.R. Smith, M. Beigi and A. Benitez, "Visual Information Retrieval from Large Distributed On-line Repositories", Communications of the ACM, December, 1997.

.... Even though some of these measures are difficult to quantify for a human observer, these global low level features have been successfully used in various content based retrieval systems to organize the contents of a database for browsing and to perform query by example (QBIC, VisualSEEk, Virage) [5]. 4.2.3 Local Structure In contrast to Global Structure, which does not provide any information about the individual parts of the image or the video sequence, the Local Structure level is concerned with the extraction and characterization of the components. At the most basic level, those ....

.... the image in Figure 6, a binary shape mask could describe the Pitcher Region and the Goalkeeper object in Figure 7 (see Figure 9) and Figure 11 (see Figure 13) Such elements have also been used in content based retrieval systems, mainly on query by usersketch interfaces such as VisualSEEk [5]. The concern here is not with objects, but rather with the basic elements that represent them and with combinations of such elements a square, for example, is formed by four lines. 4.2.4 Global Composition At this level, we focus on the specific arrangement or composition of the basic elements ....

S.-F. Chang, J. R. Smith, M. Beigi, and A. B. Benitez, "Visual Information Retrieval from Large Distributed On-line Repositories", Communications of ACM, Vol. 40, No. 12, pp. 63-71, Dec 1997. 17

No context found.

Chang, S.-F. and J. R. Smith, 1997. Visual Information Retrieval from Large Distributed Online Repositories. Communications of the ACM, 40(12): 63-71.

No context found.

S.-F. Chang, J. R. Smith, M. Beigi, and A. Benitez. Visual information retrieval from large distributed online repositories. Communications of the ACM, 40(12):63--71, Dec. 1997.

No context found.

S. Chang, J. Smith, M. Beigi & A. Benitez (1997) Visual information retrieval from large distributed online repositories. Communications of ACM 40, 63}71.

No context found.

Shih-Fu Chang, John R. Smith, Mandis Beigi and Ana Benitez, "Visual Information Retrieval From Large Distributed Online Repositories", Communications of ACM, Vol.40, No.12, 1997.

No context found.

S. Chang, J. Smith, M. Beigi, and A. Benitez, "Visual information retrieval from large distributed online repositories," Commun. ACM #, 12--20 #1997#.

No context found.

Chang S. F., Smith J. R., Beigi M., Benitez A., Visual information retrieval from large distributed online repositories , Communications of the ACM, 40 :12, pages 63-67, 1997.

No context found.

S.-F. Chang, J. Smith, M. Beigi, and A. Benitez, "Visual information retrieval from large distributed online repositories," Comm. ACM 40(12), pp. 63--71, 1997.

No context found.

S. Chang, J. Smith, M. Beigi and A. Benitez, "Visual information retrieval from large distributed online repositories", Communications of ACM, Dec. pp. 12-20, 1997.

No context found.

S. Chang, J. Smith, M. Beigi & A. Benitez 1997. Visual information retrieval from large distributed online repositories. Communications of the ACM 40(12), 63 -- 71.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC