Media Retrieval Research Articles

Self-adaptable media service architecture for guaranteeing reliable multimedia services

The main objective of this paper is to design and develop a Self-Adaptable Media Service Architecture (SAMSA) for providing reliable dynamic composite multimedia service through policy-based actions. The multimedia services such as media retrieval, transcoding, scaling and display services are combined based on the preferences of the user to create a dynamic composite multimedia service called as Video-on-Demand service. Such distributed multimedia services deployed using Service Oriented Architecture (SOA) can be accessed in heterogeneous environments that are prone to changes during run-time. To provide reliable and adaptive multimedia services, a powerful self-adaptable architecture with dynamic compositions of multimedia services is necessary to adapt during run-time and react to the environment. The adaptability in this proposed architecture is achieved by enabling the service providers to Monitor, Analyze and Act on the defined policies that support customization of compositions of multimedia services and guarantee the Quality of Service (QoS) provisioning. The Media Service Monitor (MSM) observes the business and quality metrics associated with the multimedia services during run-time. The monitored results are analyzed by Monitored Results Analyzer (MRA) which identifies the type and location of the fault. The Adaptive Media Service Manager (AMSM) takes corrective actions based on the monitored results, through the policies defined as an extension of WS-Policy (Web Service--Policy framework). The effectiveness of the proposed Self-Adaptable Media Service Architecture (SAMSA) has been evaluated on Dynamic Composite Real-time Video-on-Demand Web Service (DCRVoDWS) for a maximum of 200 simultaneous client's requests. The analysis of results shows that the proposed architecture provides better improvement on reliability, response time and user satisfaction.

Fast and accurate global motion compensation

Video understanding has attracted significant research attention in recent years, motivated by interest in video surveillance, rich media retrieval and vision-based gesture interfaces. Typical methods focus on analyzing both the appearance and motion of objects in video. However, the apparent motion induced by a moving camera can dominate the observed motion, requiring sophisticated methods for compensating for camera motion without a priori knowledge of scene characteristics. This paper introduces two new methods for global motion compensation that are both significantly faster and more accurate than state of the art approaches. The first employs RANSAC to robustly estimate global scene motion even when the scene contains significant object motion. Unlike typical RANSAC-based motion estimation work, we apply RANSAC not to the motion of tracked features but rather to a number of segments of image projections. The key insight of the second method involves reliably classifying salient points into foreground and background, based upon the entropy of a motion inconsistency measure. Extensive experiments on established datasets demonstrate that the second approach is able to remove camera-based observed motion almost completely while still preserving foreground motion.

Multi-Label Classification Method for Multimedia Tagging

Community tagging offers valuable information for media search and retrieval, but new media items are at a disadvantage. Automated tagging may populate media items with few tags, thus enabling their inclusion into search results. In this paper, a multi-label decision tree is proposed and applied to the problem of automated tagging of media data. In addition to binary labels, the proposed Iterative Split Multi-label Decision Tree (IS-MLT) is easily extended to the problem of weighted labels (such as those depicted by tag clouds). Several datasets of differing media types show the effectiveness of the proposed method relative to other multi-label and single label classifier methods and demonstrate its scalability relative to single label approaches.Keywords: Automated Multimedia Tagging; Community Tagging; Multi-label Classification; Multi-label Decision Tree; Pattern Classification

Cross Media Analysis and Retrieval on the Web: Theory and Algorithm

Cross Media Analysis and Retrieval on the Web: Theory and Algorithm

Cross Media Analysis and Retrieval on the Web: Theory and Algorithm

Cross Media Analysis and Retrieval on the Web: Theory and Algorithm

시각 미디어 온톨로지에 기반한 서비스 제공자 랭킹

인터넷 상에 산재해 있는 사진이나 비디오 등 시각 미디어 데이터를 효과적으로 검색하는 것은 전자 미술 박물관, 전자상거래, 전자 쇼핑몰 등 여러 응용 분야에서 중요한 일이다. 이러한 분야에서는 단순한 키워드 검색이 아닌 내용 기반 또는 의미 기반의 멀티미디어 검색을 필요로 한다. 인터넷 상의 시각 미디어를 효과적으로 검색하기 위해 제안된 선행 연구에서는 시각 미디어의 메타데이터와 온톨로지를 이용하고 또한 웹서비스를 이용하여 의미 기반의 검색을 수행한다. 본 연구에서는 인터넷 상에서 여러 시각 미디어 제공자와 이 제공자들의 정보를 가지고 있는 하나의 중계자가 존재하는 상황에서 시각 미디어를 효율적으로 검색하기 위한 전 단계로 적합한 서비스 제공자를 찾는 방법을 제안한다. 제안된 방법은 사용자의 질의에 적합한 제공자들과 그 순위를 효율적으로 얻기 위하여 온톨로지의 트리 구조를 이용한다. 온톨로지 트리에서 하위 노드의 크기와 자식 노드의 크기에 기반한 이 방법은 기존의 방법에 비해 효과적으로 제공자들간의 순위를 측정한다. 실험 결과 이 방법이 속도는 비슷하게 유지하면서 정확한 결과를 도출함을 보인다. It is important to retrieve effectively the visual media such as pictures and video in the internet, especially to the application areas such as electronic art museum, e-commerce, and internet shopping malls. It is also needed in these areas to have content-based or even semantic-based multimedia retrieval instead of simple keyword-based retrieval. In our earlier research, we proposed a semantic-based visual media retrieval framework for the effective retrieval of the visual media from the internet. It uses visual media metadata and ontology based on the web service to achieve the semantic-based retrieval. In this research, there are more than one visual media service providers and one central service broker. As a preliminary step to the visual media data retrieval, a method is proposed to retrieve the service providers effectively. The method uses the structure of the ontology tree to obtain the providers and their rankings. It also uses the size of sub nodes and child nodes in the tree. It measures the rankings of providers more effectively than previous method. The experimental results show the accuracy of the method while keeping compatible speed against the existing method.

Knowledge-Driven Multidimensional Indexing Structure for Biomedical Media Database Retrieval

Today, biomedical media data are being generated at rates unimaginable only years ago. Content-based retrieval of biomedical media from large databases is becoming increasingly important to clinical, research, and educational communities. In this paper, we present the recently developed entropy balanced statistical (EBS) k-d tree and its applications to biomedical media, including a high-resolution computed tomography (HRCT) lung image database and the first real-time protein tertiary structure search engine. Our index utilizes statistical properties inherent in large-scale biomedical media databases for efficient and accurate searches. By applying concepts from pattern recognition and information theory, the EBS k-d tree is built through top-down decision tree induction. Experimentation shows similarity searches against a protein structure database of 53 363 structures consistently execute in less than 8.14 ms for the top 100 most similar structures. Additionally, we have shown improved retrieval precision over adaptive and statistical k-d trees. Retrieval precision of the EBS k-d tree is 81.6% for content-based retrieval of HRCT lung images and 94.9% at 10% recall for protein structure similarity search. The EBS k-d tree has enormous potential for use in biomedical applications embedded with ground-truth knowledge and multidimensional signatures.

SEMI-AUTOMATED ANNOTATION AND RETRIEVAL OF DANCE MEDIA OBJECTS

This article considers a system for the semiautomatic annotation of an audio-visual media of dance domain, DMAR (Dance Media Annotation, authoring and Retrieval system). We present the architecture of the system, a manual annotation tool, a semiautomatic authoring tool and a search engine for the choreographers, dancers and students to demonstrate how the dance media can be semantically annotated and how this information can be used for the retrieval of the dance media objects. This article then outlines the underlying XML Schema based content description structures of DMAR and discusses the merits and demerits of our approach of evolving semantic network as the basis for the audio-visual content description. Further, this article proposes a quality metric, fidelity to evaluate the expressive power of the dance annotations. Finally, evaluation results are presented to depict the performance of the dance video queries in terms of precision and recall.

Modeling user context with applications to media retrieval

In this paper, we develop a theoretical understanding of multi-sensory knowledge and user context and their inter-relationships. This is used to develop a generic representation framework for multi-sensory knowledge and context. A representation framework for context can have a significant impact on media applications that dynamically adapt to user needs. There are three key contributions of this work: (a) theoretical analysis, (b) representation framework and (c) experimental validation. Knowledge is understood to be a dynamic set of multi-sensory facts with three key properties --- multi-sensory, emergent and dynamic. Context is the dynamic subset of knowledge that affects the communication between entities. We develop a graph based, multi-relational representation framework for knowledge, and model its temporal dynamics using a linear dynamical system. Our approach results in a stable and convergent system. We applied our representation framework to a image retrieval system with a large collection of photographs from everyday events. Our experimental validation with the retrieval evaluated against two reference algorithms indicates that our context based approach provides significant gains in real-world usage scenarios.

Information devolution?

We face enormous challenges with the phenomenal volume and diversity of information being generated. As the above quote indicates, however, in spite of our burgeoning capabilities to process, store, and retrieve information, we are just as quickly losing capability to maintain information permanence. In the strictest sense, evolution refers to gradual or episodic changes in a species, system, or process with time. Most people assume or will make the quick layperson's assertion that evolution involves changing for the better or improving with time. So do we have a word for something that is changing in a negative sense—that becomes worse or diminished in capability/utility with time? How about using “devolution” for a species, system, or process that is actually diminishing in “quality” or capability with time? I contend that the frenzied rush to a totally digital “information culture” is devolution, at least in the thrusts to totally abandon more permanent display, storage, and retrieval media. In fact, it is devolution at a pace unparalleled in recorded history—a recorded history that spans 50 000 years from cave paintings of early man to the present.

Urological diagnosis of prostate carcinoma

To summarize current urologic measures in the diagnosis of prostate cancer. Current urologic literature as listed by electronic literature search and retrieval media (PubMed, Medline) was evaluated. Current measures and trends in the diagnosis of prostate cancer were summarized. Early diagnosis of prostate cancer (PCA) is crucial because PCA can be curatively treated only in its early stages. German general health insurance covers digital rectal examination (DRE) in men over the age of 45, however measurement of prostate specific antigen-serum levels (PSA) as a screening measure for PCA is not covered. DRE, PSA-serum level measurements, transrectal ultrasound and prostate biopsy are essential in the early diagnosis of PCA.

Media Effects by Involvement Under Voluntary Exposure: A Comparison of Television, Print and Static Internet

SUMMARY In this study, the effects on consumer responses are explored of single-medium and multiple-media campaigns under two levels of consumer's product involvement. The results of this study confirm that research on media effects need to take into account consumer product involvement. Consumer involvement seems to be crucial for the differential effects of retrieval (print and Internet) and delivery (television) media. Retrieval media will be effective under conditions of high rather than low product involvement, because retrieval media have a limited capacity to influence low-involved and passive consumers. Delivery media are better suited for influencing low-involved consumers. Delivery media are able to attract the attention of low-involved consumers and, as a consequence, may induce superficial processing and low involvement learning.

QoS-based media retrieval for networked digital archives

When digital multimedia data are retrieved from distributed digital archives, Quality of Service (QoS) criteria can be used to enhance the user's usability. In this paper, assuming best-effort networks as the infrastructure, we describe a multimedia (MM) retrieval method based on the QoS concept for MM data archives in heterogeneous communication environments. We introduce a multi-level QoS model and use response time, which is calculated based on a system-level QoS criterion, as an application-level QoS criterion. The order of retrieving MM data is determined, based on both key-word matching using key words specified by the user and the application-level QoS criterion. A prototype system is developed for retrieving sightseeing-related MM data including video, images, voice, and text. The proposed method of data retrieval has a shorter response time than methods that do not take QoS criteria into account. Also, a search method based on key-word association is described that enables wide-area search and effective information finding.

Phased RGSS: An improved disk array scheduling for continuous media retrieval

This paper proposed a disk array scheduling algorithm called Phased RGSS which improves the initial service latency over RGSS presented by Kang and Park [IEICE Trans. Inform. Systems E81-D (8) (1998)] without increasing the buffer requirement for jitter-free service for continuous media retrieval. A specific data layout should be provided in order for Phased RGSS to be applied. The data layout makes disk arm movement more effective. The characteristics of Phased RGSS are proved by analysis of buffer requirement as well as initial startup delay.

Special Section on Storage, Processing, and Retrieval of Digital Media

The <i>Journal of Electronic Imaging</i> (JEI), copublished bimonthly with the Society for Imaging Science and Technology, publishes peer-reviewed papers that cover research and applications in all areas of electronic imaging science and technology.

Multimedia technologies for structuring and retrieval of TV news

Because of the media digitization, a large amount of information such as speech, audio and video data is produced everyday. In order to retrieve data from these databases quickly and precisely, multimedia technologies for structuring and retrieving of speech, audio and video data are strongly required. In this paper, we overview the multimedia technologies such as structuring and retrieval of speech, audio and video data, speaker indexing, audio summarization and cross media retrieval existing today for TV news detabase. The main purpose of structuring is to produce tables of contents and indices from audio and video data automatically. In order to make these technologies feasible, first, processing units such as words on audio data and shots on video data are extracted. On a second step, they are meaningfully integrated into topics. Furthermore, the units extracted from different types of media are integrated for higher functions.

DNA damage in round spermatids of mice with a targeted disruption of the Pp1cgamma gene and in testicular biopsies of patients with non-obstructive azoospermia.

Non-obstructive azoospermia accounts for a considerable proportion of male factor infertility. Current therapies for treatment of this kind of infertility include procedures such as intracytoplasmic sperm injection (ICSI), round spermatid injection (ROSI), round spermatid nucleus injection (ROSNI) and elongated spermatid injection (ELSI). All involve injection of haploid germ cells retrieved from testicular biopsies into recipient oocytes. We have investigated a mouse model of azoospermia for quality of haploid germ cell genomes, based on 4,6-diamidino-2-phenylindole (DAPI)/TdT-mediated dUTP nick-end labelling (TUNEL) labelling. The mouse model, a targeted mutation in the protein phosphatase 1cg gene, results in severe depletion of haploid germ cells from the round spermatid stage on. Mice homozygous for the mutation are completely infertile, and produce only the occasional spermatozoon. Spermatozoa and round spermatids retrieved from either the epididymides or the testes of mutant mice displayed very high rates of DNA fragmentation. In contrast, similar cells retrieved from heterozygous or wild-type littermates displayed low levels of DNA fragmentation. In some cases, the high rates of DNA fragmentation in mutant cells could be lowered by inclusion of antioxidants in the retrieval media. High rates of DNA fragmentation were also observed in round spermatids retrieved from testicular biospies of human patients with non-obstructive azoospermia. These results suggest that one of the features of the pathology associated with azoospermia is fragmented DNA in haploid germ cells. This raises questions about the suitability of using these cells for fertility treatment.

Dynamic scheduling techniques for interactive hypermedia servers

Future multimedia-on-demand systems will bring interactive applications in education, entertainment, advertisement, and visualisation to a large number of users at homes, offices, classrooms and laboratories. A class of multimedia data, called hypermedia, is represented by graph structures whose branches represent different video and audio segments and whose nodes represent user interactions. To provide satisfactory service in an interactive hypermedia system, disruptions and delays in the concurrent delivery of continuous media data streams from storage devices cannot be tolerated. Slow response to new requests generated by on-line users is also highly undesirable. The frequent arrivals acid completions of user requests can seriously degrade the throughput of typical servers that are designed to maintain performance at the steady state. To address these issues, we propose a dynamic approach to the retrieval of continuous media and sporadic data in an interactive hypermedia server. The throughput and efficiency of our server are improved by observing run-time information to skip or reorder service sequences, as well as maximising disk reads. Through pipelining the admission of new streams with the departure of ongoing streams, the performance is also immune to the presence of transients. User response times under identical loads are reduced significantly with respect to other existing approaches. Furthermore, our scheduler can efficiently redirect disk bandwidth to service sporadic requests when the demand on real-time services decreases. We evaluate our technique by comparing its performance with static techniques via thorough experiments under different buffer capacities, stream lengths and request arrival rates. Results of experiments show that our approach can significantly improve continuous media data throughput, user response time, as well as sporadic data throughput, and is thus a plausible approach for more interactive multimedia applications such as hypermedia.

Hierarchical image modeling for object-based media retrieval

Images are more structurally complex than the data stored in the traditional DBMSs. Many systems perform media retrieval based on image feature matching; the fact that images have structures is ignored. On the other hand, we view images as compound objects containing many component objects. Each component object corresponds to a region(s) that is a visually and semantically meaningful entity (e.g. car, man, etc.). As images/objects truly have hierarchical structures and have both visual and semantics properties, we argue that image retrieval using either semantics and image matching alone is insufficient. In this paper, we introduce a hierarchical structure for image modeling to support image retrieval using combinations of semantic expressions and visual examples at both the whole image and object levels. We give formal definitions of a multimedia query language and a system implementation based on this image modeling. We also address associated query processing issues and discuss the portability and extendibility of our approach.

An event-based model for continuous media data on heterogeneous disk servers

Over the last few years, there has been intense work on the problem of retrieval of continuous media (CM) data from disk. However, no single unified framework exists within which such retrieval problems can be studied. In this paper, we first propose a formal model for CM data retrieval from heterogeneous disk servers. This model can be used to characterize CM data retrieval problems independently of how data is laid out on disk, and what objectives (e.g., minimize client delay, maximize buffer utilization, etc.) the system manager is interested in. We then show how using this formal model, we can neatly define what it means to optimally handle events that occur in movie-on-demand (MOD) systems. Examples of such events include new clients entering the system, old clients leaving the system, continuing clients doing pause, rewind and fast-forward operations. Multiple events may occur simultaneously and we show how such events trigger state transitions in the system. We then develop an algorithm called the QuickSOL algorithm that handles events occurring in MOD systems. This algorithm works in two phases: in the first phase, it quickly finds a way of handling as many of the events occurring at time t as possible. In the second phase, it optimizes the solution found in the first phase. The advantage is that the algorithm can be interrupted anytime after the first phase is completed. We report on experiments showing that QuickSOL works well in practice.