In this paper, we present a conceptual framework for indexing different aspects of visual information. Our framework unifies concepts from this literature in diverse fields such as cognitive psychology, library sciences, art, and the more recent content-based retrieval. We present multiple level structures for visual and non-visual and non- visual information. The ten-level visual structure presented provides a systematic way of indexing images based on syntax and semantics, and includes distinctions between general concept and visual concept. We define different types of relations at different levels of the visual structure, and also use a semantic information table to summarize important aspects related to an image. While the focus is on the development of a conceptual indexing structure, our aim is also to bring together the knowledge from various fields, unifying the issues that should be considered when building a digital image library. Our analysis stresses the limitations of state of the art content-based retrieval systems and suggests areas in which improvements are necessary.
In this paper, we study computational models and techniques to combine textural and image features for classification of images on Internet. A framework is given to index images on the basis of textural, pictorial and composite information. The scheme makes use of weighted document terms and color invariant image features to obtain a high-dimensional similarity descriptor to be used as an index. Based on supervised learning, the k-nearest neighbor classifier is used to organize images into semantically meaningful groups of Internet images. Internet images are first classified into photographical and synthetical images. After classifying images into photographical and synthetical images, we further classify photographical images into portraits and non-portraits. Further, synthetical images are classified into button and non-button images.
The effective classification of the contents of an image allows us to adopt the most appropriate strategies for image enhancement, color processing, compression, and rendering. We address here the problem of distinguishing photographs from graphics and texts purely on the basis of low-level feature analysis. The preliminary results of our experimentation are reported.
With ever more popularity of video web-publishing, many popular contents are being mirrored, reformatted, modified and republished, resulting in excessive content duplication. While such redundancy provides fault tolerance for continuous availability of information, it could potentially create problems for multimedia search engines in that the search results for a given query might become repetitious, and cluttered with a large number of duplicates. As such, developing techniques for detecting similarity and duplication is important to multimedia search engines. In addition, content providers might be interested in identifying duplicates of their content for legal, contractual or other business related reasons. In this paper, we propose an efficient algorithm called video signature to detect similar video sequences for large databases such as the web. The idea is to first form a 'signature' for each video sequence by selection a small number of its frames that are most similar to a number of randomly chosen seed images. Then the similarity between any tow video sequences can be reliably estimated by comparing their respective signatures. Using this method, we achieve 85 percent recall and precision ratios on a test database of 377 video sequences. As a proof of concept, we have applied our proposed algorithm to a collection of 1800 hours of video corresponding to around 45000 clips from the web. Our results indicate that, on average, every video in our collection from the web has around five similar copies.
It is well established that humans possess cognitive abilities to process images extremely rapidly. At GTE Laboratories we have been experimenting with Web-based browsing interfaces that take advantage of this human facility. We have prototyped a number of browsing applications in different domains that offer the advantages of high interactivity and visual engagement. Our hypothesis, confirmed by user evaluations and a pilot experiment, is that many users will be drawn to interfaces that provide rapid presentation of images for browsing tasks in many contexts, among them online shopping, multimedia title selection, and people directories. In this paper we present our application prototypes using a system called PolyNav and discuss the imaging requirements for applications like these. We also raise the suggestion that if the Web industry at large standardized on an XML for meta-content that included images, then the possibility exist that rapid-fire image browsing could become a standard part of the Web experience for content selection in a variety of domains.
We explore the use of soft computing and user defined classifications in multimedia database systems for content- based queries to obtain the members of a class is a fixed set. With multimedia databases, however, an object may belong to different classes with different probabilities. In addition, alternative users may classify objects differently due to subjectivity of human perception on multimedia objects. In order to remedy for this situation, we propose a unified model that captures both conventional techniques and soft memberships. We implemented the model by extending the traditional database query capabilities such that the result of a query depends on the user who submits the query. We compared our proposed system with conventional image retrieval systems and observed a significant margin of improvement in matching the user expectations.
Increased interest in content-based storage and retrieval of images and video frames has been stemmed from its potential applications in multimedia information systems. Various matching methods have been proposed in the literature, including histogram intersection, distance method, and reference table method. A comparison of these three techniques has proved that the reference table method is the best in terms of retrieval efficiency. However, the drawback of this method is that it requires a pre-defined set of reference feature which can approximately cover all features in the selected application. While this condition may be satisfied in some applications, in situations where there are continuing additions and/or deletions to the database and where knowledge of features in the images is not available a priori, such a technique will not produce very reliable results. The reference feature or color table method requires a representative sample of all images stored in the database in order to select the reference feature or color table. For example, such a priori knowledge is impossible to obtain in a trade-marks database. To based on color-clustering, which is a computationally expensive approach. In this study, we propose an image retrieval method based on the relative entropy, known as the Kullback directed divergence. This measure is non-negative and it is zero if and only if two distributions are identical; i.e., perfect match. Erel has only one minimum for every comparison. This offers a unique criterion for optimization with low computational complexity.It also provides a thoughtful view for the type of data distribution in the sense that the whole range of data distribution is considered in matching and not only some moments.
Nearest Neighbor search is a fundamental task in many applications. At present state of the art, approaches to nearest neighbor search are not efficient in high dimensions. In this paper we present an efficient angle based balanced index structure called AB-tree, which uses heuristics to decide whether or not to access a node in the index tree based on the estimated angle and the weight of the node. We have presented the result of AB-tree for up to 64 dimensions, and have shown that the performance of the AB-tree algorithm does not deteriorate when processing nearest neighbor queries as dimension increases. However, it is no longer guaranteed that the entire true K nearest neighbors to a query point will be found. Extensive experiments on synthetic data and real data demonstrate that the search time is improved by a factor of up to 85 times over that of SS-tree in 64 dimension while maintaining 90 percent accuracy. Real data includes color histogram and corner like features in a heterogenous collection of natural images.
Spatial distribution of color is very important for refining color histograms use din indexing and retrieving color images. Existing histogram refinement techniques are based on the spatial distribution of a single color or color pair. In this paper, the concept of spatial distribution of a subset of colors, which is defined as the occurrence of different geometric configurations of the color set, is used to provide new clues for refining traditional color histogram. The concept is a unification of some existing techniques such as color density maps, color correlogram and color tuples. Experimental results demonstrate that triangular geometric histogram, on e of the simplest special cases of geometric histograms, which is defined as the occurrence of a list of isosceles right triangles of different side lengths of color triples, is more desirable than existing techniques for content-based image retrieval, especially when the database in question consists of on-line color images which are extremely heterogenous in terms of the content of images, camera types, lighting conditions and so on.
One of the requirements for the fast growing technology of multimedia and Internet is image retrieval. A retrieval scheme needs to be efficient, and effective in finding similar images. This requires a robust retrieval scheme against rotation, reflection, translation, scaling, illumination and noise with low computational cost. In this paper a new scheme which overcomes the problems of previous retrieval systems such as sensitivity to illumination, false edges, translation, rotation, noise is introduced. The computational cost of this method is comparable to the previous methods. In this new scheme the image edges will be extracted first, then the edge angles are quantized. Based on correlation between amplitude and phase of neighboring edges the edge orientation correlogram, which is a 2D matrix, is generated. This matrix is normalized and ordered in such a wy that it becomes invariant to rotation, reflection, scaling and translation. This matrix can be used as a feature vector for describing the image and also as an index in image databases. The experimental result shows this new method is superior to other color-based, color-spatial and shape-based indexing schemes.
The Web provides a large repository of multimedia data, text, images, etc. Most current search engines focus on textural retrieval. In this paper, we focus on using an integrated textural and visual search engine for Web documents. We support query refinement which proves useful and enables cross-media browsing in addition to regular search.
In this paper we address the issue of scheme change detection on MPEG encoded video sequences with the use of combined video and audio information. We present the architecture of a system which provides an integration framework for algorithms handling both kinds of information and we show how these can be combined in order to provide a suitable segmentation of the video content. Finally, we discuss the first steps for a distributed version of the proposed architecture.
We have developed a file format which is well-suited for network applications invovling images of documents. It is extremely compact and flexible, using the emerging JBIG2 standard and the Mixed Raster Content imaging model. DigiPaper is designed for ease of document storage and interchange: it shares image elements across multiple pages, can be created from a wide range of sources and can be decompressed easily. The data structures within DigiPaper also map well onto page description languages such as PostScript or PDF, enabling efficient printing using current printer architectures. We describe some of the algorithms used in DigiPaper for JBIG2 compression of black and white scanned documents.
Recent work in the Internet Engineering Task Force (IETF) has focused on the transport of image file formats over the Internet that typically represent documents suitable for output on paper or other graphically rich display terminals. This paper describes the recent work on the Internet Printing Protocol and Internet Fax. The protocols, image file formats selected for inclusion and the rationale for the various engineering choices made are discussed. A considerable number of new products and services support these new protocols and it is the belief of the authors that these technologies represent the principal means by which document images will be transported over the Internet in the future. New work within the IETF is concentrate on extending the protocols and principals embodied in both schemes to create a more robust 'quality document distribution' mod.
Synchronized Multimedia Integration Language (SMIL) is a recommendation developed by the W3C Synchronized Multimedia Working Group. SMIL is a declarative authoring language to specify multimedia synchronized presentation using a time line. We present the design and implementation of a domain specific SMIL Authoring Tool (SAT). The output generate from SAT adheres to the standard as the presentation is specified in a human-readable and machine-friendly XML-tagged file. SAT can be use din conjunction with any SMIL-compliant players. It provides a high-level point-and-click tool, which can create and update the logics of SMIL presentations. It is also designed to provide template-based authoring, which is deemed to be easy-to-create for domain- specific presentations.
Some improvements of embedded zerotree wavelet algorithm are considere. Compression methods tested here are based on dyadic wavelet image decomposition, scalar quantization and coding in progressive fashion. Profitable coders with embedded form of code and rate fixing abilities like Shapiro EZW and Said nad Pearlman SPIHT are modified to improve compression efficiency. We explore the modifications of the initial threshold value, reconstruction levels and quantization scheme in SPIHT algorithm. Additionally, we present the result of the best filter bank selection. The most efficient biorthogonal filter banks are tested. Significant efficiency improvement of SPIHT coder was finally noticed even up to 0.9dB of PSNR in some cases. Because of the problems with optimization of quantization scheme in embedded coder we propose another solution: adaptive threshold selection of wavelet coefficients in progressive coding scheme. Two versions of this coder are tested: progressive in quality and resolution. As a result, improved compression effectiveness is achieved - close to 1.3 dB in comparison to SPIHT for image Barbara. All proposed algorithms are optimized automatically and are not time-consuming. But sometimes the most efficient solution must be found in iterative way. Final results are competitive across the most efficient wavelet coders.
The indices obtained from tree-structured vector quantization have two capabilities. First, they can provide image with different resolutions which gives an hierarchical order image based on the closeness of image blocks. Second, image blocks indices, depending on the tree depth, give the characteristics of neighboring pixels of an image. These indices characteristics have ben used in generating a feature vector which shows the image clusters in different resolutions with the capability of giving information about the neighboring pixels characteristics including edges or smooth image areas. This method has been compared with other previous image retrieval scheme based on vector quantization.
We present a JAVA-based Interactive Progressive Local Image Transmission (IPLIT) syste for viewing large images over the bandwidth-limited WWW in 'reasonable time'. One motivation behind this research is the need for medical specialists to remotely view medical imags, in reasonable time, over the WWW. In our IPLIT system, the user employs a JAVA-based Internet browser to view and browse a low resolution image. The identification of features or regions of interest before observing those regions in detail is performed by either selecting a particular region manually via mouse or by utilizing an automatic feature-detection mode. The automatic feature-detection displays high-resolution subimages along a trajectory determined by the user-specified feature of interest. Our program handles 3D image data as a sequence of 2D images. Our IPLIT system is tested on actual MRI, CT and Ultrasound medical images obtained from the Robarts Research Institute at the University of Western Ontario, Canada. One such image was used as the test image in this paper. A few test images were borrowed from the Human Visual Project.
This paper presents a multiresolution approach for progressive image coding at very-low bit rate and transmission via web. A real application concerning a real- time traffic monitoring system for unattended motorway or urban areas is considered. First, semantic information, e.g., alarm generation and vehicle classification provided by a recognition subsystem, is sent to a remote control center; then, binary shape information is transmitted, in order to allow the remote operator to validate the alarm situation. Pictorial information can be also required as a further step by the operator of the control center. The coding scheme allows to use the web as useful transmission channel.
Display color can be represented by a simple two-stage model, although the display performance is affected by aging, magnetic field sand various other factors. For optimum image rendering there should be a close match between the source and destination devices, in terms of primary chromaticities, white point, gamma, palette encoding and gamut mapping. Characterization procedures for displays include visual assessment, visual matching and measurement with a tricolorimeter or telespectroradiometer. Color may be communicated in images either by encoding the image directly in a standard color space, or by attaching a profile that allows the data values to be interpreted. Color correction of Internet images may be carried out on either the receiver side or on the server side. Evaluation of the effectiveness of color Internet image delivery should include repeatability and consistency of the display characterization procedure, color accuracy of the imagery and general usability considerations.
In this study, 34 students and teachers form a college of fine arts selected suitable colors for their artistic work on CRT display. From the results, several fundamental colors selected by artists can be seen on CIE-xy color spaces. The result of our experiment clearly showed that the 34 subjects could be divided into three groups based on differences in the level of red light or green light in their color palette.
This study examined the new 3D phase of seeing from the perception of depth on the CRT screen with stereoscopic vision. The experiment accomplished with attaching a stereoscope to the CRT display. The stimulus presented on the CRT screen is a pair of the rectangles placed in leftside to be vertical and rightside to be horizontal position against a white background. The two chromatic colors were presented to each rectangle. 12 types were recognized on the perception of transparency and dominance on the figure of superimposed rectangles. Subjects are asked to observe the stimulus and report the phase of the intersection of vertical and horizontal rectangles, for example, either vertical or horizontal stimulus is dominant over the other.
Visual retrieval by content in an Image DataBase (IDB) is still an open problem. So far, various methods with different semantic levels have been developed, for internet search or off-line IDBs, but few of them take into account the user's perceptual point of view. Two features primarily used for visual retrieval in IDBs are shape and color. We focus our attention on color, from the perspective of color appearance. The Human Visual System (HVS) has adaptation mechanisms that cause the user to perceive the relative chromaticity of an area, rather than its absolute color. In addition, due to the acquisition process, color distortions are added to heterogeneous IDBs. Digital pictures of real objects for IDBs must be digitized and the acquisition process is composed of various passages and means, each one introducing unwanted color shifting. Moreover, the color quantization and the device gamut can introduce additional distortion on the original color information. The overall result is a recognizable and the device gamut can introduce additional distortion on the original color information. The overall result is a digital image that can significantly differ in color from the real object. For the user the image may still be easily recognizable, but the color search change can vary widely and differ for each image or for the same image with different acquisition processes. For this reason, the user's perceptual point of view must be added into the management of color. The idea presented in this paper adds a pre-filtering algorithm that simulates the HVS and that discounts the acquisition color distortion in the query image as well as in each image in the IDB. Moreover, we suggest to use for the image retrieve, a more perceptively linear chromatic distance in the color comparison.
Comparison in the RGB domain is not suitable for precise color matching, due to the strong dependency of this domain on factors like spectral power distribution of the light source and object geometry. We have studied the use of multispectral or hyperspectral images for color matching, since it can be proven that hyperspectral images can be made independent of the light source and object geometry. Hyperspectral images have the disadvantages that they are large compared to regular RGB-imags, which makes it infeasible to use them for image matching across the Internet. For red roses, it is possible to reduce the large number of bands of the spectral images to only three bands, the same numbers of an RGB-image, using Principal Component Analysis, while maintaining 99 percent of the original variation. The obtained PCA-images of the roses can be matched using for example histogram cross correlation. From the principal coordinates plot, obtained from the histogram similarity matrices of twenty images of red roses, the discriminating power seems to be better for normalized spectral images than for color constant spectral images and RGB-images, the latter being recorded under highly optimized standard conditions.
A Japanese office environment is done in the big room, and a room is limited to the officer. A fluorescent light is used directly as a lighting in the big room from the ceiling. Light from the outside is easy to be reflected on the CRT screen because it follows and a partition isn't set up in the big room. A Japanese has the habit which a ceiling is hated to bedark than. A degree of diffusion is low though many lights of prevention of reflection were developed. A ceiling gets dark, and cleaning is difficult as for the reason. As for this research. 1) The desk arrangement that the Internet is easy to build, and design dimension. 2) It thought that the Japanese paper which was Japanese traditional material was used. Japanese paper has the character which makes beautiful dispersion light. It was reflected to the CRT screen, and Japanese paper paid attention to making the indoor environment of the Internet which fatigue is rare in. 3) The size of the furniture the Internet is easy to operate as to the old man, and the screen.4) The desk arrangement of decreasing influences on the surrounding environment of the electromagnetic wave as much as possible. When a special cap is put between the tables which it meets to secure wiring of the cable, wiring for the Internet and depth are made at a low price in the arrangement of the computer machine as that result as for being short of the depth. There is effect which makes light to penetrate soft in the Japanese shoji. Thickness influences the quality of the material of the paper to make an excellent indoor environment. Because the number of colors recognized when it ages decreased rapidly, it was found out that you had better be little as for the color in screen.
Hamming claimed 'the purpose of computing is insight, not numbers.' In a variant of that aphorism, we show how the Virtual Reality Modeling Language (VRML) can provide powerful insight into the mathematical properties of numbers. The mathematical problem we consider is the relatively recent conjecture colloquially known as the '3x + 1 problem'. It refers to an iterative integer function that also can be though of as a digraph rooted at unity with the other numbers in any iteration sequence locate at seemingly randomized positions throughout the tree. The mathematical conjecture states that there is a unique cycle at unity. So far, a proof for this otherwise simple function has remained intractable. Many difficult problems in number theory, however, have been cracked with the aid of geometrical representations. Here, we show that any arbitrary portion of the 3x + 1 digraph can be constructed by iterative application of a unique subgraph called the G-cell generator - similar in concept to a fractal geometry generator. We describe the G-cell generator and present some examples of the VRML worlds developed programmatically with it. Perhaps surprisingly, this seem to be one of the few attempts to apply VRML to problems in number theory.
The possibility of associating an easy-to-use mental 3D model with a computer interface is the main innovation of the Trini Diagram. For the first time, data of an emotional character can be positioned in relation to one another, permitting an intersubjective interpretation and treatment using computer tools. One of the potential applications for this new technique is that of the 'interfacing engine' for cataloging and retrieval in image banks. The Trini Diagram can also become a fundamental architecture for the construction of 'subjective interfaces' for a new form of man-machine interaction.
Hypervideo is the natural evolution of Hypertext. Interlinking images and text in modern Hypertext pages is well understood and widely used in commercial services. Links from text to images and image to text are used interchangeably and there is a plethora of development environments in the market today.Links from video sequences out to other pieces of information are not currently widely available mainly because of the temporal nature of any defined hotspots. For example in a video commercial about cars a hotspot defined around a car model featured in the commercial has to 'move' with or rather track the car. The obvious way of defining such a hotspot is to use a mixture of conventional image mapping techniques assisted by manual mapping of the required regions in every frame. This approach is very cumbersoem and time consuming so will never have widespread commercial appeal.
In developmental biology an overwhelming amount of experimental data concerning patterns of gene expression is produced revealing the genetic layout of the embryo and finding evidence for anomalies. Genes are part of complex genetic cascades and consequently their study requires tools for handling combinatorial problems. Gene expression is spatio-temporal and generally, imagin is used to analyze expression in four dimensions. Reporting and retrieving experimental data has become so complex that printed literature is no longer adequate and therefore databases are being implemented. Zebrafish is a popular model system in developmental biology. We are developing a 3D digital atlas of the zebrafish embryo, which is envisaged as standard allowing comparisons of experimentally induced and normally developing embryos. This 3D atlas is based on microscopical anatomy. From serial sections 3D images are reconstructed by capturing section images and registering these images respectively. This is accomplished for al developmental stages. Data management is solved using XML which is platform independent, ASCII-based, interchangeable and allows easy browsing. Applying supervised segmentation accomplishes a completely anatomically annotated 3D image. It divides the image into domains required for comparison and mapping. Experts provided with dedicated software and Internet-access to the images review annotations. Complete annotation and review is stored in a database.
The Visible Human Slice Server started offering its slicing services at the end of June 1998. From that date until the end of May, more than 280,000 slices were extracted from the Visible Man, by layman interested in anatomy, by students and by specialists. The Slice Server is based one Bi-Pentium PC and 16 disks. It is a scaled down version of a powerful parallel server comprising 5 Bi-Pentium Pro PCs and 60 disks. The parallel server program was created thanks to a computer-aided parallelization framework, which takes over the task of creating a multi-threaded pipelined parallel program from a high-level parallel program description. On the full blown architecture, the parallel program enables the extraction and resampling of up to 5 color slices per second. Extracting 5 slice/s requires to access the disks and extract subvolumes of the Visible Human at an aggregate throughput of 105 MB/s. The publicly accessible server enables to extract slices having any orientation. The slice position and orientation can either be specified for each slice separately or as a position and orientation offered by a Java applet and possible future improvements. In the very near future, the Web Slice Server will offer additional services, such as the possibility to extract ruled surfaces and to extract animations incorporating slices perpendicular to a user defined trajectory.
TextOscillatorFont is the name of a package containing a wave shaped font or computer typeface. This package links image and information in a single language and is designed to be used as a tool without constraints.
Increasing numbers of research libraries are turning to the Internet for electron interlibrary loan and for document delivery to patrons. This has been made possible through the widespread adoption of software such as Ariel and DocView. Ariel, a product of the Research Libraries Group, converts paper-based documents to monochrome bitmapped images, and delivers them over the Internet. The National Library of Medicine's DocView is primarily designed for library patrons are beginning to reap the benefits of this new technology, barriers exist, e.g., differences in image file format, that lead to difficulties in the use of library document information. To research how to overcome such barriers, the Communications Engineering Branch of the Lister Hill National Center for Biomedical Communications, an R and D division of NLM, has developed a web site called the DocMorph Server. This is part of an ongoing intramural R and D program in document imaging that has spanned many aspects of electronic document conversion and preservation, Internet document transmission and document usage. The DocMorph Server Web site is designed to fill two roles. First, in a role that will benefit both libraries and their patrons, it allows Internet users to upload scanned image files for conversion to alternative formats, thereby enabling wider delivery and easier usage of library document information. Second, the DocMorph Server provides the design team an active test bed for evaluating the effectiveness and utility of new document image processing algorithms and functions, so that they may be evaluated for possible inclusion in other image processing software products being developed at NLM or elsewhere. This paper describes the design of the prototype DocMorph Server and the image processing functions being implemented on it.
This paper present the results of the activities concerning the architecture of a high quality catalogue for made-to- measure garments undertaken in the ISHTAR project. The syste is based upon servers and standard Internet browsers. The main issues faced in this paper are the definition of an architecture to obtain the best trade off between performance and representation of apparel/textile goods and the introduction of a color management system that can benefit also generic internet users.
This paper present a method to distribute the security key used to encrypt digital content by embedding the security key as a watermark into a digital sample that is appended to the encrypted digital content. The security key embedded into the sample is imperceptible and can not be detected by attackers. To play or read digital content, a portable device will extract the security key from the sample, then use the detected key to decrypt the encrypted content. This paper also present a method to generate and transfer right between clearing hose and portable device and between two portable devices. In order to prevent digital content from being illegally used, authentication, and verification between clearing house and portable device and between two portable devices are employed. The right information is tamper-resistent when it is transferred to portable devices. This method can maximally protect the usage of digital contents even if the portable device is breached.
In the framework of M.Cube Project 22266, 1996-1999, ESPRIT Program of E.U., an informative system called Mediterranean Cultural Directory - MCD, has been developed to support cultural initiatives and to promote the European Cultural Heritage. The major objective of M.Cube is the provision of services to multimedia developers and users of cultural multimedia, both directly and through inter-mediation.
Foreign language materials on the web are growing at a faster rate than English language materials and it has been predicted that by the end of 1999 the amount of non-English resources on the internet will exceed English resources. A significant portion of the non-English material is in the form of images.
Digital watermarking has been indicated as a technique in the position to cope with the problem of Intellectual Property Rights (IPR) protection of images; this result should be achieved by embedding into the data an unperceivable digital code, namely the watermark, carrying information about the copyright status of the work to be protected. In this paper, the practical feasibility of IPR protection through digital watermarking is investigated. The most common requirements application scenarios impose to the watermarking technology to satisfy are discussed. Watermarking schemes are first classified according to the approach used to extract the embedded code and then the impact, such a classification has on watermark usability, is investigated form an application point of view. As it will be shown, the effectiveness of watermarking as an IPR protection tool turns out to be heavily affected by the detection strategy, which as to be carefully matched to the application at hand. Finally, the practical case of the Tuscany and Gifu Art Virtual Gallery has been considered in detail, to further explain in which manner a watermarking technique can be actually used.
The Visible Human Data set is an important contribution to the national collection of anatomical images. To enhance the availability of these images, the National Library of Medicine has supported the design and development of a prototype object-oriented image database which imports, stores, and distributes high resolution anatomical images in both pixel and voxel formats. One of the key database modules is its client-server Internet interface. This Web interface provides a query engine with retrieval access to high-resolution anatomical images that range in size from 100KB for browser viewable rendered images, to 1GB for anatomical structures in voxel file formats. The Web query and retrieval client-server system is composed of applet GUIs, servlets, and RMI application modules which communicate with each other to allow users to query for specific anatomical structures, and retrieve image data as well as associated anatomical images from the database. Selected images can be downloaded individually as single files via HTTP or downloaded in batch-mode over the Internet to the user's machine through an applet that uses Netscape's Object Signing mechanism. The image database uses ObjectDesign's object-oriented DBMS, ObjectStore that has a Java interface. The query and retrieval systems has been tested with a Java-CDE window system, and on the x86 architecture using Windows NT 4.0. This paper describes the Java applet client search engine that queries the database; the Java client module that enables users to view anatomical images online; the Java application server interface to the database which organizes data returned to the user, and its distribution engine that allow users to download image files individually and/or in batch-mode.