Audio-Visual Communications & Networks (AVCN)

Research and development capacity in image and video compression standards for applications ranging from mobile video to high definition digital television. The standards we have expertise in include JPEG, JPEG-LS, JPEG2000, MPEG-2, MPEG-4, H263, H264 AVC, H264 SVC, H264 MVC and distributed video coding.

The thrust of R&D is concentrated on mainly three directions: 

Improving the operational system and user factors of visual compression standards through parameter optimisation and modelling. These factors include compression efficiency, statistical/visual quality metrics, real time performance, memory usage etc.

Customisation of the standards to particular application and user requirements and improvement of their respective performance. Such applications include digital photography for entertainment, medical imaging, enabling of user interactivity in visual compression standards, personalisation of users’ visual experience, video for mobile devices, reducing the quality fluctuation among TV commercials etc.

Provision of added value services on visual compression standards such as editing capabilities, video summarisation, video analysis, visual content description using computer vision etc.

There is a lot of experience in algorithmic development and simulation of standard based software image/video coders(JPEG,JPEG-LS,JPEG2000,MPEG-2,MPEG-4, H263,H264 AVC, H264 SVC, H264 MVC, MPEG-7, distributed video coding etc.)

Developed Optimisations focus on a variety of aspects such as:

Application dependent information loss distributions in JPEG-LS. (e.g. digital photography, medical imaging etc.)

Design of the first rate control algorithm for JPEG-LS.

Computationally affordable RD optimisations in MPEG-2.

Compression savings without sacrificing quality in MPEG-4 VOPs

ME speed ups in MPEG-2 through predictive scanning orders and computation distribution

H264X(X is AVC, SVC, MVC) related optimisations

Distributed video coding

 

Fig.1 Visual quality improvement through rate control

Areas of research focuses on the creation and manipulation of 3D soundfields based upon Ambisonics. The theory of Ambisonics and the development of Ambisonic recording and reproduction techniques. Application of Ambisonics to performance enhancement and teaching.

Software development of encoders and decoders – development of a suite of plug-ins for use in DAW.

Theoretical investigations into soundfield recording.

Manipulation of soundfields by computer techniques.

Applications relate to performing enhancement for musicians and creation of immersive sound fields for games technology.

Research and development capacity in various wireless and mobile networks and applications:

Multi-homing support: Distribute and redistribute different live application flows adaptively over multiple network interfaces based on dynamic application-specific policies in multi-homed mobile networks based on NEMO, SCTP etc.

Mobility management: Cross-layer protocols to smoothly switch ongoing sessions between subnets (handover management) and track mobile users and terminals (location management) through optimisation and integration of Mobile IP, SIP etc.

Wireless sensor networks: Energy-aware proactive MAC, routing and mobility –enhanced schemes to extend the network’s lifetime

Wireless mesh networks: Instant mesh networking with scalable multicast and efficient routing for emergency management in incident sites.

 

Fig.2 Flexible delivery and redistribution of applications over multiple wireless access networks simultaneously

The STB Tester platform developed by AIMS (INT) Ltd. is designed to test the transmission of Internet Protocol TV (IPTV) systems and evaluate the perceived video quality in a lab setting. The project investigates methods and tools towards extending the testing capabilities and facilities available in the Tester and thus enabling the product to meet the demanding market requirements. Based on a realistic IPTV testbed, the project explores a range of technical issues such as video transmission and evaluation in constrained or impaired network conditions and speeding up the current time-consuming visual quality evaluation process.

 

Project for TileLite Ltd.

Development of software application facility at GMI. To develop and embed within the company a formal design capability for PC based software applications.

SCoRE Project with LEDTEK LTd

To develop and assess a range of high efficiency LED drivers for industrial applications.

MEDUSA (Multi Environment Deployable Universal Software Application)

“MEDUSA is concerned with the identification of situations associated with gun related threats, based on behavioural interpretation of CCTV data and through combining psychological and image processing approaches.

The project aims to develop a new machine learning system for the detection of individuals carrying guns through the use of CCTV surveillance networks. The system will combine both human and machine-based factors in achieving this.” (From MEDUSA website)

MULTINET (Intelligent service delivery over multiple networks)
“The MULTINET communication system will be capable of providing the network and application functionalities so that multiple simultaneous networks can be seamlessly handled to optimise communications in multiple dimensions; while sustaining the existing wireless and mobile industry and attracting new business revenue.

The priority and focus will be on multi-homing intelligent user PC terminals, providing intelligent network selection and dynamic bearer modifications between WLAN-WLAN initially and progressing to WLAN-UMTS / WLAN-WiMAX.” (From MULTINET website)

DIAS (Design, Implementation, Adaptation of Sensor Networks)
“The overall goal of the project is to develop methods and tools for the design, implementation, and adaptation of entire environmental sensor network systems. Each resulting sensor network is optimal with respect to a global cost function specified by the network designer.

We will focus on sensor networks that consist of small, battery-powered, wireless sensors that may be tethered or mobile. For such systems, we need to minimize power consumption while delivering full functionality.” (From DIAS website)

CrimeVis (Improving the usefulness of CCTV for Crime Detection through automated annotation of video streams and selective compression)

Implementation of an optical design facility at VMS.
To embed within the company a certified optical test and design facility for a range of LED signage products.

Maximising the End-to-End quality of the MPEG-4 & the H26L video coding standards over wired networks
“Maximising End-to-End video quality in wired networks is a very desirable feature from the user's point of view. However, the extent to which this is achievable depends not only on the video coding scheme but also on the network conditions and the decoder specifications. This holistic view of the video coding and transport problem reveals the need for derivation of optimal and nearly optimal parameter sets (profiles) which determine End-to-End quality by taking into account the influence of the coding system, the condition of the network and the decoder specifications. Profile generation covering a large gamut of parameters for single (quality) or multi-objective optimisations is always a bottleneck in compression standards such as MPEG-4 or the emerging H26L, since it is done on a trial and error basis and by using empirical evidence rather than sound mathematical theory. As such, the derived profiles can not represent optimal or nearly optimal settings for the interacting parameters in a video coding and transport system. This research proposal aims at maximising End-to-End video quality in a single and multi-objective optimisation framework by using iterative techniques both in the encoder-decoder and the transcoder-decoder scenarios and also in developing simple two way protocols that are able to re-optimise profiles dynamically, based on changing network conditions and availability of decoder resources.” (From EPSRC website)

PARADIGM – Project for Applied Research in Advanced Digital Manufacturing

Mobility Support Architectures of Next-Generation IP-Based Wireless Networks

Next-generation wireless systems are expected to support efficient IP-based terminal and personal mobility when a user is on the move within a single geographical or administrative domain or across domains. Novel mobility-support architectures were designed to this end. The focus of this project was the optimisation and integration of Mobile IP (and its variants) and Session Initiation Protocol (SIP), which dynamically combines and leverages the advantages of both protocols in a cost-effective way. The design motivation and methodology is to make full use of standardised work from both protocols, select composite processes that are more efficient for common functions, integrate similar entities and procedures to reduce redundancies, and avoid further duplicate standardisation. By these means, the system efficiency can be greatly improved.