Search

Transnational education is one of the many facets of inclusive education, with its relevance of English as the natural Lingua Franca in a vast range of contexts. The University of Catania, Italy, is an interesting case-study of a highly international University, which welcomes students from all over the world, and where transnational education has become an ordinary practice. In the light of my experience of English language teaching, I would like to report both on the cognitive value of an intercultural, learner-centered approach and on the various factors that concur to make mixed classes a motivating practice. Based on the hypothesis that people who speak different languages have different worldviews, my teaching proposal aims at fulfilling two objectives: understanding the perspectives of other cultures, and comparing the language and culture of others with students’ own.

West Texas west of the one-hundredth meridian has experienced a commoditized form of agriculture for less than two hundred years. In that period of time two major agricultural industries have implemented themselves in an arid land hoping the wide- open land would provide their industries with ample territory for their endeavors to prosper. The annual rainfall fell below the twenty-inch mark deemed to be the dividing line between assisted and non-assisted agriculture. It was because of the arid nature of the region that settlers deemed water, and access to it, to be the most important commodity. The push of the Texas cattle industry past the ninety-eighth and one-hundredth meridians intensified the search for any means of water production to avoid altering the industry itself. The beginning-of-the-end of water sources came with the implementation of the windmills in west Texas. By the turn of the twentieth century, a more creative way to produce rainfall came in the form of dynamite. The detonation of dynamite was used in west Texas to attempt to compel the sky to relinquish moisture over the barren cow pastures and cotton fields to relieve the effects of the arid landscape. The continued migration of cotton farmers and new pumping technologies being produced by the end of the Second World War further depleted underground water supplies. A realistic attempt (compared to dynamiting for rain) to aid in dealing with the effects of aridity, organizations began to incorporate to channel water from rivers and streams to convey water to west Texas and eastern New Mexico. These were the first attempts to supplement the limited water supply west of the one-hundredth meridian and continued the mass production of agricultural commodities for the market economy.

Effective indexing is crucial for providing convenient access to scanned versions of large collections of historically valuable handwritten manuscripts. Since traditional handwriting recognizers based on optical character recognition (OCR) do not perform well on historical documents, recently a holistic word recognition approach has gained in popularity as an attractive and more straightforward solution (Lavrenko et al. in proc. document Image Analysis for Libraries (DIAL’04), pp. 278–287, 2004). Such techniques attempt to recognize words based on scalar and profile-based features extracted from whole word images. In this paper, we propose a new approach to holistic word recognition for historical handwritten manuscripts based on matching word contours instead of whole images or word profiles. The new method consists of robust extraction of closed word contours and the application of an elastic contour matching technique proposed originally for general shapes (Adamek and O’Connor in IEEE Trans Circuits Syst Video Technol 5:2004). We demonstrate that multiscale contour-based descriptors can effectively capture intrinsic word features avoiding any segmentation of words into smaller subunits. Our experiments show a recognition accuracy of 83%, which considerably exceeds the performance of other systems reported in the literature.

Video retrieval is mostly based on using text from dialogue and this remains the most signi¯cant component, despite progress in other aspects. One problem with this is when a searcher wants to locate video based on what is appearing in the video rather than what is being spoken about. Alternatives such as automatically-detected features and image-based keyframe matching can be used, though these still need further improvement in quality. One other modality for video retrieval is based on segmenting objects from video and allowing end users to use these as part of querying. This uses similarity between query objects and objects from video, and in theory allows retrieval based on what is actually appearing on-screen. The main hurdles to greater use of this are the overhead of object segmentation on large amounts of video and the issue of whether we can actually achieve effective object-based retrieval. We describe a system to support object-based video retrieval where a user selects example video objects as part of the query. During a search a user builds up a set of these which are matched against objects previously segmented from a video library. This match is based on MPEG-7 Dominant Colour, Shape Compaction and Texture Browsing descriptors. We use a user-driven semi-automated segmentation process to segment the video archive which is very accurate and is faster than conventional video annotation.

In this paper, an integrated approach to fast and efficient construction of statistical shape models is proposed that is a potentially useful tool in Information Retrieval(IR). The tool allows intuitive extraction of accurate contour examples from a set of images using a semi-automatic segmentation approach. The user is allowed to draw on the scene by simply dragging a mouse over the image and creating a set of labelled scribbles for the objects to be segmented. An automatic segmentation algorithm uses the scribbles to partition the scene and extract objects’ contour. A set of labelled points (landmarks) is identified automatically on the set of examples thereby allowing statistical modeling of the objects’ shape. The main contribution of this paper is the new approach to automatic landmark identification eliminating the burden of manual landmarking. The approach utilizes a robust method for pairwise correspondence proposed originally in [1, 2]. The landmarks are used to train statistical shape models known as Point Distribution Models (PDM) [11]. Qualitative results are presented for 3 classes of shape which exhibit various types of nonrigid deformation.

This paper addresses the issue of classification of lowlevel features into high-level semantic concepts for the purpose of semantic annotation of consumer photographs. We adopt a multi-scale approach that relies on edge detection to extract an edge orientation-based feature description of the image, and apply an SVM learning technique to infer the presence of a dominant building object in a general purpose collection of digital photographs. The approach exploits prior knowledge on the image context through an assumption that all input images are �outdoor�, i.e. indoor/outdoor classification (the context determination stage) has been performed. The proposed approach is validated on a diverse dataset of 1720 images and its performance compared with that of the MPEG-7 edge histogram descriptor.

The creation of personalised 3D characters has evolved to provide a high degree of realism in both appearance and animation. Further to the creation of generic characters the capabilities exist to create a personalised character from images of an individual. This provides the possibility of immersing an individual into a virtual world. Feature detection, particularly on the face, can be used to greatly enhance the realism of the model. To address this innovative contour based templates are used to extract an individual from four orthogonal views providing localisation of the face. Then adaptive facial feature extraction from multiple views is used to enhance the realism of the model.

The holy grail of multimedia indexing and retrieval is developing algorithms capable of imitating human abilities in distinguishing and recognising semantic concepts within the content, so that retrieval can be based on ”real world” concepts that come naturally to users. In this paper, we discuss an approach to using segmented video objects as the midlevel connection between low-level features and semantic concept description. In this paper, we consider a video object as a particular instance of a semantic concept and we model the semantic concept as an average representation of its instances. A system supporting object-based search through a test corpus is presented that allows matching presegmented objects based on automatically extracted lowlevel features. In the system, relevance feedback is employed to drive the learning of the semantic model during a regular search process.

This paper presents a robust and efficient method for segmentation of images into large regions that reflect the real world objects present in the scene. We propose an extension to the well known Recursive Shortest Spanning Tree (RSST) algorithm based on a new color model and so-called syntactic features [1]. We introduce practical solutions, integrated within the RSST framework, to structure analysis based on the shape and spatial configuration of image regions. We demonstrate that syntactic features provide a reliable basis for region merging criteria which prevent formation of regions spanning more than one semantic object, thereby significantly improving the perceptual quality of the output segmentation. Experiments indicate that the proposed features are generic in nature and allow satisfactory segmentation of real world images from various sources without adjustment to algorithm parameters.

In this paper, we describe our experiments for TRECVID 2004 for the Search task. In the interactive search task, we developed two versions of a video search/browse system based on the Físchlár Digital Video System: one with text- and image-based searching (System A); the other with only image (System B). These two systems produced eight interactive runs. In addition we submitted ten fully automatic supplemental runs and two manual runs. A.1, Submitted Runs: • DCUTREC13a_{1,3,5,7} for System A, four interactive runs based on text and image evidence. • DCUTREC13b_{2,4,6,8} for System B, also four interactive runs based on image evidence alone. • DCUTV2004_9, a manual run based on filtering faces from an underlying text search engine for certain queries. • DCUTV2004_10, a manual run based on manually generated queries processed automatically. • DCU_AUTOLM{1,2,3,4,5,6,7}, seven fully automatic runs based on language models operating over ASR text transcripts and visual features. • DCUauto_{01,02,03}, three fully automatic runs based on exploring the benefits of multiple sources of text evidence and automatic query expansion. A.2, In the interactive experiment it was confirmed that text and image based retrieval outperforms an image-only system. In the fully automatic runs, DCUauto_{01,02,03}, it was found that integrating ASR, CC and OCR text into the text ranking outperforms using ASR text alone. Furthermore, applying automatic query expansion to the initial results of ASR, CC, OCR text further increases performance (MAP), though not at high rank positions. For the language model-based fully automatic runs, DCU_AUTOLM{1,2,3,4,5,6,7}, we found that interpolated language models perform marginally better than other tested language models and that combining image and textual (ASR) evidence was found to marginally increase performance (MAP) over textual models alone. For our two manual runs we found that employing a face filter disimproved MAP when compared to employing textual evidence alone and that manually generated textual queries improved MAP over fully automatic runs, though the improvement was marginal. A.3, Our conclusions from our fully automatic text based runs suggest that integrating ASR, CC and OCR text into the retrieval mechanism boost retrieval performance over ASR alone. In addition, a text-only Language Modelling approach such as DCU_AUTOLM1 will outperform our best conventional text search system. From our interactive runs we conclude that textual evidence is an important lever for locating relevant content quickly, but that image evidence, if used by experienced users can aid retrieval performance. A.4, We learned that incorporating multiple text sources improves over ASR alone and that an LM approach which integrates shot text, neighbouring shots and entire video contents provides even better retrieval performance. These findings will influence how we integrate textual evidence into future Video IR systems. It was also found that a system based on image evidence alone can perform reasonably and given good query images can aid retrieval performance.

This paper describes an approach for detecting dialogue scenes in movies. The approach uses automatically extracted low- and mid-level visual features that characterise the visual content of individual shots, and which are then combined using a state transition machine that models the shot-level temporal characteristics of the scene under investigation. The choice of visual features used is motivated by a consideration of formal film syntax. The system is designed so that the analysis may be applied in order to detect different types of scenes, although in this paper we focus on dialogue sequences as these are the most prevalent scenes in the movies considered to date.

In this paper we present an algorithm for automatic classification of sound into speech, instrumental sound/ music and silence. The method is based on thresholding of features derived from the modulation envelope of the frequency limited audio signal. Four characteristics are examined for discrimination: the occurrence and duration of energy peaks, rhythmic content and the level of harmonic content. The proposed algorithm allows classification directly on MPEG-1 audio bitstreams. The performance of the classifier was evaluated on TRECVID test data. The test results are above-average among all TREC participants. The approaches adopted by other research groups participating in TREC are also discussed.

In this paper we propose a novel audio-visual feature-based framework, for event detection in field sports broadcast video. The system is evaluated via a case-study involving MPEG encoded soccer video. Specifically, the evidence gathered by various feature detectors is combined by means of a learning algorithm (a support vector machine), which infers the occurrence of an event, based on a model generated during a training phase, utilizing a corpus of 25 hours of content. The system is evaluated using 25 hours of separate test content. Following an evaluation of results obtained, it is shown for this case, that both high precision and recall statistics are achievable.

A significant challenge in large multimedia databases is the provision of efficient means for semantic indexing and retrieval of visual information. Artificial text in video is normally generated in order to supplement or summarise the visual content and thus is an important carrier of information that is highly relevant to the content of the video. As such, it is a potential ready-to-use source of semantic information. In this paper we present an algorithm for detection and localisation of artificial text in video using a horizontal difference magnitude measure and morphological processing. The result of character segmentation, based on a modified version of the Wolf-Jolion algorithm [1][2] is enhanced using smoothing and multiple binarisation. The output text is input to an “off-the-shelf” noncommercial OCR. Detection, localisation and recognition results for a 20min long MPEG-1 encoded television programme are presented.

The “Físchlár” systems are a family of tools for capturing, analysis, indexing, browsing, searching and summarisation of digital video information. Físchlár-News-Stories, described in this paper, is one of those systems, and provides access to a growing archive of broadcast TV news. Físchlár-News-Stories has several notable features including the fact that it automatically records TV news and segments a broadcast news program into stories, eliminating advertisements and credits at the start/end of the broadcast. Físchlár-News-Stories supports access to individual stories via calendar lookup, text search through closed captions, automatically-generated links between related stories, and personalised access using a personalisation and recommender system based on collaborative filtering. Access to individual news stories is supported either by browsing keyframes with synchronised closed captions, or by playback of the recorded video. One strength of the Físchlár-News-Stories system is that it is actually used, in practice, daily, to access news. Several aspects of the Físchlár systems have been published before, bit in this paper we give a summary of the Físchlár-News-Stories system in operation by following a scenario in which it is used and also outlining how the underlying system realises the functions it offers.

In this paper, we describe our experiments for both the News Story Segmentation task and Interactive Search task for TRECVID 2003. Our News Story Segmentation task involved the use of a Support Vector Machine (SVM) to combine evidence from audio-visual analysis tools in order to generate a listing of news stories from a given news programme. Our Search task experiment compared a video retrieval system based on text, image and relevance feedback with a text-only video retrieval system in order to identify which was more effective. In order to do so we developed two variations of our Físchlár video retrieval system and conducted user testing in a controlled lab environment. In this paper we outline our work on both of these two tasks.

This paper describes a method for automatically detecting human faces in generic video sequences. We employ an iterative algorithm in order to give a confidence measure for the presence or absence of faces within video shots. Skin colour filtering is carried out on a selected number of frames per video shot, followed by the application of shape and size heuristics. Finally, the remaining candidate regions are normalized and projected into an eigenspace, the reconstruction error being the measure of confidence for presence/absence of face. Following this, the confidence score for the entire video shot is calculated. In order to cluster extracted faces into a set of face classes, we employ an incremental procedure using a PCA-based dissimilarity measure in con-junction with spatio-temporal correlation. Experiments were carried out on a representative broadcast news test corpus.

In this paper we present an overview of the software tools currently available for the creation and display of MPEG-4 content. We first describe tools for encoding raw video into MPEG-4 compliant bitstreams. We then describe how this content may be used to create a complete MPEG-4 scene containing both graphical and interactive elements in addition to the more usual video and audio elements. Clearly, MPEG-4 content cannot be viewed without appropriate decoders and players and these are addressed in the third section of this paper. Finally, we demonstrate how these tools may be combined together to create MPEG-4 applications by presenting the details of two sample applications we have developed.

Large volumes of information in video format are being created and made available from a number of application areas, including movies, broadcast TV, CCTV, education video materials, and so on. As this information is increasingly in digital format, this creates the opportunity and then the demand for content-based access to such material. One particular kind of video information that we are interested in is broadcast TV news and in this paper we report on our work on developing content-based access to broadcast TV news. Our work is carried out within the context of the Físchlár system, developed to allow content access to large volumes of digital video information. We report our work on Físchlár-News which provides text search based on closed caption information as well as our on-going work on segmenting TV News programmes and providing personalised intelligent access to TV news stories, on fixed as well as mobile platforms.

The imminent rapid expansion in the number of TV channels is driving the need for efficient digital video indexing, browsing and playback systems. For the past four years, the Centre for Digital Video Processing in DCU has been working towards the provision of such a system. The current stage of development is demonstrated on our Web-based digital video system called Físchlár [1,2], which is now in hourly use by over 1000 registered users. At present a user can pre-set the recording of TV broadcast programmes and can choose from a set of different browser interfaces which allow navigation through the recorded programmes. As our research has developed we have been adding increased functionality such as personalisation and programme recommendation, automatic recording, SMS/WAP/PDA alerting, searching, summarising and so on. There have been many approaches to automatic highlight detection and summarization of sports programmes [3-6]. However, most methods are oriented to particular sports and/or specific edit effects or need much computational effort. Therefore we decided to take a more basic and generic approach based on the observation that, in most common sports coverage with a commentator and spectators, exciting events are usually characterized by an increase in the audio amplitude, particularly in the speech band. Measurement of the audio amplitude is described in Section II and a case study is presented in Section III. Section IV summarises results while conclusions and areas of continuing and future work are outlined in Section V.

Dublin City University participated in the Feature Extraction task and the Search task of the TREC-2002 Video Track. In the Feature Extraction task, we submitted 3 features: Face, Speech, and Music. In the Search task, we developed an interactive video retrieval system, which incorporated the 40 hours of the video search test collection and supported user searching using our own feature extraction data along with the donated feature data and ASR transcript from other Video Track groups. This video retrieval system allows a user to specify a query based on the 10 features and ASR transcript, and the query result is a ranked list of videos that can be further browsed at the shot level. To evaluate the usefulness of the feature-based query, we have developed a second system interface that provides only ASR transcript-based querying, and we conducted an experiment with 12 test users to compare these 2 systems. Results were submitted to NIST and we are currently conducting further analysis of user performance with these 2 systems.

This paper gives an overview of low-power techniques proposed in the literature for mobile multimedia and Internet applications. Exploitable aspects are discussed in the behavior of different video compression tools. These power-efficient solutions are then classified by synthesis domain and level of abstraction. As this paper is meant to be a starting point for further research in the area, a lowpower hardware & software co-design methodology is outlined in the end as a possible scenario for video-codec-on-a-chip implementations on future mobile multimedia platforms.

In this paper, we present the MPEG-1 Audio bitstreams processing work which our research group is involved in. This work is primarily based on the processing of the encoded bitstream, and the extraction of useful audio features for the purposes of analysis and browsing. In order to prepare for the discussion of these features, the MPEG-1 audio bitstream format is first described. The Application Interface Protocol (API) which we have been developing in C++ is then introduced, before completing the paper with a discussion on audio feature extraction.

In today’s fast paced world, the time available to watch long sports programmes is decreasing, while the number of sports channels is rapidly increasing. Many viewers desire the facility to watch just the highlights of sports events. This paper presents a simple, but effective, method for generating sports video highlights summaries. Our method detects semantically important events in sports programmes by using the Scale Factors in the MPEG audio bitstream to generate an audio amplitude profile of the program. The Scale Factors for the subbands corresponding to the voice bandwidth give a strong indication of the level of commentator and/or spectator excitement. When periods of sustained high audio amplitude have been detected and ranked, the corresponding video shots may be concatenated to produce a summary of the program highlights. Our method uses only the Scale Factor information that is directly accessible from the MPEG bitstream, without any decoding, leading to highly efficient computation. It is also rather more generic than many existing techniques, being particularly suitable for the more popular sports televised in Ireland such as soccer, Gaelic football, hurling, rugby, horse racing and motor racing.

The Centre for Digital Video Processing at Dublin City University developed Fischlar [1], a web-based system for recording, analysis, browsing and playback of digitally captured television programs. One major issue for Fischlar is the automatic evaluation of video quality in order to avoid processing and storage of corrupted data. In this paper we propose an automatic classification technique that detects the video content quality in order to provide a decision criterion for the processing and storage stages.

This paper describes a proposed algorithm for speech/music discrimination, which works on data directly taken from MPEG encoded bitstream thus avoiding the computationally difficult decoding-encoding process. The method is based on thresholding of features derived from the modulation envelope of the frequency-limited audio signal. The discriminator is tested on more than 2 hours of audio data, which contain clean and noisy speech from several speakers and a variety of music content. The discriminator is able to work in real time and despite its simplicity, results are very promising.

The Físchlár digital video system is a web-based system for recording, analysis, browsing and playback of TV programmes which currently has about 350 users. Although the user interface to the system is designed for desktop PCs with a large screen and a mouse, we are developing versions to allow the use of mobile devices to access the system to record and browse the video content. In this paper, the design of a PDA user interface to video content browsing is considered. We use a design framework we have developed previously to be able to specify various video browsing interface styles thus making it possible to design for all potential users and their various environments. We can then apply this to the particulars of the PDA's small, touch-sensitive screen and the mobile environment where it will be used. The resultant video browsing interfaces have highly interactive interfaces yet are simple, which requires relatively less visual attention and focusing, and can be comfortably used in a mobile situation to browse the available video contents. To date we have developed and tested such interfaces on a Revo PDA, and are in the process of developing others.

In this paper we describe a structured approach for the development of user interfaces for the Fíchlár video browsing system, a web-based system for recording, browsing and playback of TV programmes. The user interface to the system was originally designed for desktop use with a large screen and a mouse and we are currently developing versions suitable for mobile device (PDA) access to the system. We review a design framework for video browsing interface formats and some of the formats developed for desktop and PDA use, including interfaces for the Psion Revo and Compaq iPAQ PDAs. This work is driven by the need to investigate how best to include the user in the content specification and retrieval loop and how to and the various balance points between user interaction and system automation.

This paper describes the organisational and playback features of Físchlár, a digital video library that allows users to record, browse and watch television programmes online. Programmes that can be watched and recorded are organised by personal recommendations, genre classifications, name and other attributes for access by general television users. Motivations and interactions of users with online television libraries are outlined and they are also supported by personalised library access, categorised programmes, a combined player browser with content viewing history and content marks. The combined player browser supports a user who watches a programme on different occasions in a non-sequential order.

The development of standards for video encoding coupled with the increased power of computing mean that content-based manipulation of digital video information is now feasible. Shots are a basic structural building block of digital video and the boundaries between shots need to be determined automatically to allow for content-based manipulation. A shot can be thought of as continuous images from one camera at a time. In this paper we examine a variety of automatic techniques for shot boundary detection that we have implemented and evaluated on a baseline of 720,000 frames (8 hours) of broadcast television. This extends our previous work on evaluating a single technique based on comparing colour histograms. A description of each of our three methods currently working is given along with how they are evaluated. It is found that although the different methods have about the same order of magnitude in terms of effectiveness, different shot boundaries are detected by the different methods. We then look at combining the three shot boundary detection methods to produce one output result and the benefits in accuracy and performance that this brought to our system. Each of the methods were changed from using a static threshold value for three unconnected methods to one using three dynamic threshold values for one connected method. In a final summing up we look at the future directions for this work.

In digital video indexing research area an important technique is called shot boundary detection which automatically segments long video material into camera shots using content-based analysis of video. We have been working on developing various shot boundary detection and representative frame selection techniques to automatically index encoded video stream and provide the end users with video browsing/navigation feature. In this paper we describe a demonstrator digital video system that allows the user to record a TV broadcast programme to MPEG-1 file format and to easily browse and playback the file content online. The system incorporates the shot boundary detection and representative frame selection techniques we have developed and has become a full-featured digital video system that not only demonstrates any further techniques we will develop, but also obtains users’ video browsing behaviour. At the moment the system has a real-user base of about a hundred people and we are closely monitoring how they use the video browsing/navigation feature which the system provides.

The application of image processing techniques to achieve substantial compression in digital video is one of the reasons why computer-supported video processing and digital TV are now becoming commonplace. The encoding formats used for video, such as the MPEG family of standards, have been developed primarily to achieve high compression rates, but now that this has been achieved, effort is being concentrated on other, content-based activities. MPEG-7, for example is a standard intended to support such developments. In the work described here, we are developing and deploying techniques to support content-based navigation and browsing through digital video (broadcast TV) archives. Fundamental to this is being able to automatically structure video into shots and scenes. In this paper we report our progress on developing a variety of approaches to automatic shot boundary detection in MPEG-1 video, and their evaluation on a large test suite of 8 hours of broadcast TV. Our work to date indicates that different techniques work well for different shot transition types and that a combination of techniques may yield the most accurate segmentation.

The central issues of this dissertation are (a) what should we be doing — what problems should we be trying to solve — in order to build computer vision systems, and (b) what relevance biological vision has to the solution of these problems. The approach taken to tackle these issues centres mostly on the clarification and use of information-based ideas, and an investigation into the nature of the processes underlying perception. The primary objective is to demonstrate that information theory and extensions of it, and measurement theory are powerful tools in helping to find solutions to these problems. The quantitative meaning of information is examined, from its origins in physical theories, through Shannon information theory, Gabor representations and codes towards semantic interpretations of the term. Also the application of information theory to the understanding of the developmental and functional properties of biological visual systems is discussed. This includes a review of the current state of knowledge of the architecture and function of the early visual pathways, particularly the retina, and a discussion of the possible coding functions of cortical neurons. The nature of perception is discussed from a number of points of view: the types and function of explanation of perceptual systems and how these relate to the operation of the system; the role of the observer in describing perceptual functions in other systems or organisms; the status and role of objectivist and representational viewpoints in understanding vision; the philosophical basis of perception; the relationship between pattern recognition and perception, and the interpretation of perception in terms of a theory of measurement These two threads of research, information theory and measurement theory are brought together in an overview and reinterpretation of the cortical role in mammalian vision. Finally the application of some of the coding and recognition concepts to industrial inspection problems are described. The nature of the coding processes used are unusual in that coded images are used as the input for a simple neural network classifier, rather than a heuristic feature set The relationship between the Karhunen-Loève transform and the singular value decomposition is clarified as background the coding technique used to code the images. This coding technique has also been used to code long sequences of moving images to investigate the possibilities of recognition of people on the basis of their gait or posture and this application is briefly described.

Object-based video representations are considered to be useful for easing the process of multimedia content production and enhancing user interactivity in multimedia productions. Object-based video presents several new technical challenges, however. Firstly, as with conventional video representations, compression of the video data is a requirement. For object-based representations, it is necessary to compress the shape of each video object as it moves in time. This amounts to the compression of moving binary images. This is achieved by the use of a technique called context-based arithmetic encoding. The technique is utilised by applying it to rectangular pixel blocks and as such it is consistent with the standard tools of video compression. The blockbased application also facilitates well the exploitation of temporal redundancy in the sequence of binary shapes. For the first time, context-based arithmetic encoding is used in conjunction with motion compensation to provide inter-frame compression. The method, described in this thesis, has been thoroughly tested throughout the MPEG-4 core experiment process and due to favourable results, it has been adopted as part of the MPEG-4 video standard. The second challenge lies in the acquisition of the video objects. Under normal conditions, a video sequence is captured as a sequence of frames and there is no inherent information about what objects are in the sequence, not to mention information relating to the shape of each object. Some means for segmenting semantic objects from general video sequences is required. For this purpose, several image analysis tools may be of help and in particular, it is believed that video object tracking algorithms will be important. A new tracking algorithm is developed based on piecewise polynomial motion representations and statistical estimation tools, e.g. the expectationmaximisation method and the minimum description length principle.