IEEE MultiMedia

PrePrint: An Online Video Recommendation Framework Using View Based Tag Cloud Aggregation

Recommender systems make use of various sources of information about users such as preferences and usage patterns, and provide personalized recommendations, facilitating the information retrieval and filtering processes. In particular, with the recent unprecedented growth of user generated content in so-cial media sites, the role of recommender systems that can prune large information spaces and assist users in discovering items becomes increasingly important. This paper presents a simple, yet powerful recommendation framework that attempts to utilize the characteristics unique to the problem of recom-mending online videos annotated with tags. The proposed method operates by constructing a per-user profile as an aggregate of the tag clouds of videos viewed by the user, and recommendations are then generated based on the viewing patterns of similar users identified according to a similarity function over the user profiles. Experimental results demonstrate that the tag clouds aggregated for a user can effectively characterize the user profile for similarity computation, enabling more accurate recommenda-tions particularly for the casual users with a few video views in the social media communities that support tagging.

PrePrint: Integral and Networked Home Automation Solution towards Indoor Ambient Intelligence

We have been listening for a long time to the wide functionality that home automation technologies can offer for improving our lives. However the price and an unstable domotics market have restricted the deployment of such systems. Nowadays, companies offer too technology-dependent solutions which do not cover user demands completely. Meanwhile, works in the literature focus on small innovations on specific parts of home automation systems, which do not consider integration and deployment issues in order to present practical designs. The system presented in this work considers user requirements, including novel advances, all in an integral home automation solution suitable for many services. The modular nature of the architecture allows direct adaptation to specific cases using standard domotic technologies, for managing in-house devices, and a proposal of an IP-based network for connecting the main home automation module with the rest of platform elements. A remote security system has been developed and managing tasks are enabled by in-home control panels and an advanced 3D application for local/remote homeowner access. The system has been deployed on a prototype house, where a wide set of domotic services have been tested. Moreover, the range of indoor pervasive applications has also been extended to eHealth, elderly adaptation, greenhouse automation and energy efficiency.

PrePrint: GeoDec: A Framework to Effectively Visualize and Query Geospatial Data for Decision-Making

In this paper, we discuss GeoDec, our end-to-end system that enables geospatial decision-making by virtualizing the real-world geolocations. With GeoDec, first the geolocation of interest is rapidly and realistically simulated and all relevant geospatial data are accurately fused and embedded in the virtualized model. Subsequently, users can interactively formulate abstract decision-making queries in terms of a wide range of fundamental spatiotemporal queries supported by GeoDec and evaluate the queries in order to verify decisions in the virtual world prior to executing the decisions in real world. GeoDec blends a variety of techniques developed in the fields of databases, artificial intelligence, computer graphics and computer vision into an integrated three-tier architecture. We elaborate on various components of this architecture, which includes an extensive, multimodal and dynamic data tier, an efficient, expressive and extensible query-interface tier and an immersive, flexible and effective presentation tier.

PrePrint: Memory-spot: A technology for labeling

We present memory-spot, as a complementary labeling technology for the Internet of Things along with RFID, barcodes and other pervasive technologies. Unlike traditional RFID and barcodes, the battery-less memory-spot allows read/write/append of mega-bytes of digital content at an on-air data rate of 10 Mbps. As well as providing physical objects with locally-stored digital content, memory-spot can also contain links to web-based content in the same way as RFID. Enabling applications synthesized from local and remote data components provides both compelling utility and an excellent user experience. Several of these applications are discussed in the paper. Additionally, having a significant amount of locally accessible content significantly reduces the possibility of service degradations arising from latencies and inaccessibility of the wireless access network and back-end servers. The paper also addresses the underpinnings of the technology and describes the innovations in its antenna, modem design, processor and security engine.

PrePrint: Practical Web-based Smart Spaces

Mobile devices are evolving into hubs of content and context information. There have been many research projects showing the potential for new applications for pervasive computing. We aim to support pervasive applications on a wide variety of devices using Web and resource-based smart spaces. We address several issues for adapting a resource-based style of HTTP (REST) for pervasive services to enable easy mashup of applications in this environment. First, for security and access control in heterogeneous, dynamic environments we introduce a flexible access control mechanism on top of OpenID and OAuth. Additionally to support finding resources we use a search engine that can collaborate with existing service and network discovery mechanisms. We also outline how an emerging W3C standard, DCCI, can be used to share information within a device in an interoperable fashion.

PrePrint: Multimedia Presentation for Computer Games and Web 3.0

The HTML standard is an old and text-based format that is designed for narrowband networks. New enhancements to HTML, as well as new formats, are being evaluated. We wish to present a multimedia presentation system, format, and use cases for a wide range of context uses, including Web 3.0 and computer games. The system is compact and enables high performance 2D and 3D graphics. Use cases include in-game heads-up displays, remote playing of games, and 3D authoring for Web 3.0.

PrePrint: 3-D Reconstruction of the Human Ribcage Based on Chest X-Ray Images and Geometric Template Models

In order to avoid later stages of lung diseases, preventative methods such as regular preemptive screenings are an absolute necessity. For these screenings, costly methods, such as CT scans, are often times cost prohibitive. X-ray imagery is significantly more cost effective. However, it does not provide a three-dimensional view of the patient’s torso. Hence, this paper describes a technique that, based on a series of geometric template models and two X-ray images of a subject’s chest, creates a patient specific 3-D reconstruction of the ribcage. The 3-D reconstruction helps improve visualization of the chest without needing to use expensive 3-D scanners and can be combined with existing computer-aided disease detection techniques to help with diagnosis.

PrePrint: Pervasive Computing in Daidalos

This paper provides a summary of the research undertaken in the area of pervasive computing in the European FP6 IST project Daidalos II. Pervasiveness is one of the five key concepts underlying the Daidalos project. We provide an overview and motivation for our research, we justify and describe the overall architecture, and we discuss some of the core functional components.

PrePrint: Using Modality Replacement to Facilitate Communication Between Blind and Hearing Impaired People

The present paper presents a multimodal framework for the intercommunication between blind and hearing impaired people. The algorithms that are developed are based on the concept of situated modality replacement, which is the use of information originating from various modalities to compensate for the missing input modality of the system or the users. The proposed framework features sign language recognition and synthesis, speech recognition and synthesis and haptic interaction in virtual environments, all integrated in a virtual treasure hunting game, where the blind and the hearing-impaired users have to collaborate so as to navigate in the virtual environment and solve the riddle of the game. The treasure hunting game serves both as a multimodal entertainment tool and as an educational game training and familiarizing the users with new assistive technologies. Usability evaluation of the system has shown that the framework has significant potential for disabled users.

IEEE MultiMedia - January March 2010 (Vol. 17, No. 1)

IEEE MultiMedia

PrePrint: An Alignment Approach for Context Prediction Tasks in UbiComp environments

We study the prediction accuracy of algorithms for context prediction in ubiquitous computing environments. The discussion details the alignment prediction approach, an accurate time series estimation technique applicable to numeric and non-numeric data alike. In simulations conducted on real data sets the alignment prediction algorithm shows promising results compared to ARMA, Markov, PCA and ICA prediction approaches.

Presented By:

NEC

Ads by Pheedo

PrePrint: Visual Navigation for Mobile Devices

We present the integration of improved camera pose recovery into our landmark-based visual navigation system for mobile devices. We use detailed camera pose information in order to leverage existing images in a database, allowing us to render novel views without the addition of new images. New views are synthesized by either using a simple warping method or by utilizing automatically generated 3D models. A live client localizes and augments user-captured images by matching against a database of existing images. Camera pose information is then used to project landmark regions onto the images, allowing the highlight, selection, and retrieval of additional information on landmarks. Finally, we analyzed the performance of our system in terms of speed, accuracy and usability: our quantitative system measurements reveal accuracy and response times acceptable for interactive applications, and our user study suggest our improvements do indeed add value to the navigation application.

PrePrint: Space Matters: Physical-Digital and Physical-Virtual Co-Design in the inSpace Project

We provide an overview of the methods and products of the inSpace project, emphasizing the impact of physical-digital co-design on our creation of a meeting space and a range of physical-digital artifacts, and its relevance to the design of hybrid physical-virtual spaces that represent the current phase of our project.

PrePrint: Ubiquitous Advertising: The Killer Application for the 21st Century

This survey article proposes the idea that advertising is the next major application for ubiquitous computing. Part of the support for this idea is that ubiquitous computing applications will very likely be supported by advertising, continuing the success of advertising on the Worldwide Web and paralleling the predicted growth of mobile advertising. More interesting, however, is that ubiquitous computing will eventually support advertising in several ways. We explain how advertisers are already adopting certain ubiquitous computing technologies, and we show how ubiquitous computing research can help advertisers in the areas of ad targeting, ad feedback, customer awareness, and privacy. The article concludes with a scenario illustrating ubiquitous advertising along with suggestions for ubiquitous computing researchers in light of our predictions.

PrePrint: Reclaiming Our Shared Public Spaces – Bridging the Virtual and Physical through Social Networking and Media

For many of us, social networking and media sharing applications and services have dominated the ways of communicating and sharing our lives in Internet with friends, family and colleagues during last years. These sites carry large amount of digital media and information about who we really are; content and online profiles giving clues about our personalities. This information can be used in conjunction with modern technology to create dynamic, entertaining and useful applications that bridge the physical and online worlds. We present Social Surroundings – an application drawing its power from content already residing in online services and natural everyday places such as cafes and pubs. We introduce how to utilize the digital self-presentations people have created online to reveal just a little more - to hopefully improve and encourage natural communication and lower the social barrier of good old chatting and meeting face to face in public.

PrePrint: Detection of Risk Factors of a Toddler’s Fall Injuries

Thorough and consistent supervision is required to prevent accidental injuries of a toddler who learns and practices motor skills lacking in judgment about risks, but it is not practical. The vision-based system proposed here detects three risk factors of a toddler’s fall in the home environment in order to aid the supervision of a caregiver who stays nearby a toddler but cannot always give thorough attention to the toddler. The risk factors have been identified by analyzing official suggestions of organizations for child safety and records of toddler fall events at home. The crucial technical challenge was to differentiate a toddler, human, from other foreground objects in images. We exploited multiple dynamic motion cues for the human detection differently from most of previous studies, which use cues related to human appearance.

PrePrint: Cooperative Relative Positioning

Many pervasive applications deal with relative positions between interacting entities rather than global coordinates. The RELATE project has developed different sensing methods and a modular system architecture for peer-to-peer relative positioning, and studied these in application case studies on mobile spatial interaction, firefighter navigation, and wearable activity recognition.

PrePrint: Linking digital media to physical documents: Comparing content-based and marker-based tags

Paper is static but it is also light, flexible, robust, and has high resolution for reading documents in various scenarios. Digital devices will likely never match the flexibility of paper, but come with all of the benefits of computation and networking. Tags provide a simple means of bridging the gap between the two media to get the most out of both. In this paper, we explore the tradeoffs between two different types of tagging technologies – marker-based and content-based – through the lens of four systems we have developed and evaluated at our lab. From our experiences, we extrapolate issues for designers to consider when developing systems that transition between paper and digital content in a variety of different scenarios.

PrePrint: Example-Based Objective Quality Estimation for Compressed Images

Quantization noise is one of the dominant distortions of image compression, and its amplitude usually needs to be estimated for image quality assessment, restoration and enhancement. One such estimation, the peak signal-to-noise ratio (PSNR), has commonly been used as an objective quality measure. However, this measure has limitation in practical applications as it requires as a reference the original image, which is not always available to end users. To overcome the limitation, blind or non-reference PSNR estimation has received much attention in the literature as it requires not the original image, but some statistics of the original image, such as the probability density functions (PDFs) of original discrete cosine transform (DCT) coefficients. Assuming that PDFs of DCT coefficients follow Laplacian distribution, we propose here a new method to estimate the key parameter of the distribution from a set of training data, consisting of a variety of typical images compressed with various quantization parameters. Our experimental results show that the proposed method can estimate the PSNR of a given image more accurately, with smaller estimation bias and variance, as compared to the existing methods.

PrePrint: Understanding Recording Technologies in Everyday Life

Electronic recording and surveillance systems are arguably some of the most pervasive technologies in the world today. Despite this rapid proliferation and their study by many researchers, there is still work to be done in understanding how people reason about these technologies when they encounter them. In this paper, we describe attitudes, perceptions, and concerns regarding electronic recording encountered in daily activities. We present data gathered from interviews grounded in real experiences by the day reconstruction method. These data form the basis of a discussion for how people develop mental models about the intent and uses of a broad scope of recording technologies embedded in the world. Individual constructions of reality about current recording systems, including the people, places, and activities that surround them, provide insight into how design, technology, and policy can work together to provide appropriate information about the existence and uses of recording devices. These insights can lead to usable systems that allow individual end users to make informed personal decisions about recording in everyday experiences.

PrePrint: An MPEG-7 Compatible Video Retrieval System with Integrated Support for Complex Multimodal Queries

We present BilVideo-7, an MPEG-7 compatible, video indexing and retrieval system that supports complex multimodal queries in a unified framework. An MPEG-7 profile is developed to represent the videos by decomposing them into Shots, Keyframes, Still Regions and Moving Regions. The MPEG-7 compatible XML representations of videos according to this profile are obtained by the MPEG-7 compatible video feature extraction and annotation tool of BilVideo-7, and stored in a native XML database. Users can formulate text-based semantic, color, texture, shape, location, motion and spatio-temporal queries on an intuitive, easy-to-use Visual Query Interface, whose Composite Query Interface can be used to specify very complex queries containing any type and number of video segments with their descriptors. The multi-threaded Query Processing Server parses incoming queries into subqueries and executes each subquery in a separate thread. Then, it fuses subquery results in a bottom-up manner to obtain the final query result. The whole system is unique in that it provides very powerful querying capabilities with a wide range of descriptors and multimodal query processing in an MPEG-7 compatible interoperable environment. We present sample queries to demonstrate the capabilities of the system.

PrePrint: Layer 5 Session Handoff: Mobility Management for Video Streaming Sessions among Heterogeneous Networks and Devices

In the past decade, mature video processing and Internet technologies have made video services over IP feasible and increasingly popular, e.g., video-on-demand (VoD) and IPTV. Considering mobile life over heterogeneous networks, a user can exploit any device to enjoy/share pictures, videos and music at anytime and anywhere, and he may change his currently used device or attached network to another one. To manage video streaming sessions among heterogeneous networks and devices, we developed a Ubiquitous Multimedia Service Platform for Mobile Life over the Digital Home Environment (UMS-MLDH) in a 3-year (2006-2009) National Telecommunication Project (NTP). Many distinct-domainsˇ core techniques are devised and well integrated in UMS-MLDH, including session mobility management, session signaling and communication and adaptive layered video streaming over a 3-tier server-proxy-client network architecture. In our experiments, the system integration, implementation and merits of UMS-MLDH are exhibited over a hybrid Home-Vehicular-Office network.

PrePrint: Cooperative Beam-forming in "Smart Dust": Getting rid of Multi-hop Communications

Bridging the digital with the physical world is made feasible through a proposed system architecture utilizing recent developments in cooperative beam-forming, RFIC technology and spread-spectrum techniques. In the proposed system, nodes will have to be activated only when they are triggered to send their own data to the fusion centre, without burdening the network with MAC layer collisions and energy-consuming routing algorithms, thus guaranteeing long lifetime cycles. Due to the simplicity of the node architecture that is used to implement the proposed system, the cost, size and energy consumption of the nodes can finally justify the use of the term ‘smart dust’.

PrePrint: The Case for VM-based Cloudlets in Mobile Computing

Resource poverty is a fundamental constraint that severely limits the class of applications that can be run on mobile devices. We present a vision of mobile computing that breaks free of this fundamental constraint. In this vision, mobile users seamlessly utilize nearby computers to obtain the resource benefits of cloud computing without incurring WAN delays and jitter. Rather than relying on a distant "cloud," a mobile user instantiates a "cloudlet" on nearby infrastructure and uses it via a wireless LAN. Crisp interactive response for immersive applications that augment human cognition is then much easier to achieve because of the proximity of the cloudlet. We confirm that a critical aspect of this vision, namely rapid customization of cloudlet infrastructure, is achievable through dynamic VM synthesis.