Skip to main content

Methods for Optimisation and Subjective Assessment in Image Communications

Objective

The main objective of the MOSAIC project is to develop and prove reliable methods for the assessment of picture quality and degradation. It will also investigate operational system features which influence subjective image quality such as: audio quality, relative audio/video ruggedness and delay, and system re-synchronisation time. Associated developments in statistics will also be undertaken, both for the purposes of picture content analysis and also for the processing and presentation of results.
Assessment of picture quality and degradation have been carried out. The methods employed were largely those of the psychologist, interpreted in the light of statistical analyses. Experiments have revealed clear evidence of the scene dependent and time dependent variations when video is coded at low bit rates. A forgetfulness effect has been identified, in conjunction with a time averaging effect when a transitory impairment is embedded in a long section of unimpaired video. As a consequence, a single stimulus assessment methodology is being developed and refined. Experiments have been conducted to assess the acceptable threshold for the recovery time of digital codecs (ie the period of interruption acceptable to viewers when channel hopping or when a failure is caused by errors in the transmission path). A first evaluation of acceptable audio video delay has indicated larger tolerances than anticipated. Preliminary conclusions from tests to determine the preferred viewing distance from the screen have indicated surprisingly small differences when conventional 625 line television displays were compared with presentations of high definition television (HDTV).
Technical Approach

The methods employed in MOSAIC are largely those of the psychologist, based around a number of carefully designed experiments. The results need to be skilfully interpreted in the light of statistical analyses. It is only after novel ideas have been experimented with and refined, that new methods for the assessment of picture quality will emerge. In this way the project will:

- Develop accurate tools to specify, optimise and evaluate image communication systems.
- Support other European projects with an assessment methodology for the evaluation of image compression systems.
- Contribute towards new ITU picture quality and degradation assessment procedures.

Key Issues

- Assessment Methodology.
- Assessment of Scene Dependent Quality.
- Influence of Audio on Subjective Video Quality.
- System Parameters and Viewing Conditions.

Expected Impact

The new tools developed by the MOSAIC project will allow the developers of image transmission systems to make scientifically-based evaluations of their systems and to compare in a meaningful manner the relative performance of the various systems.
This work will be disseminated through a handbook and a workshop in autumn 1995.

Coordinator

Independent Television Commission
Address
Kings Worthy Court
SO23 7QA Winchester
United Kingdom

Participants (7)

Centre Commun d'Études de Télécommunications et de Télédiffusion (CCETT)
France
Address
4 Rue Du Clos Courtel
35512 Cesson-sévigne
EUROPEAN BROADCASTING UNION
Switzerland
Address
Ancienne Route 17A
1218 Geneva
MERCURE
France
Address
La Caltais
35890 Bourg-les-comptes
NEDERLANDSE PHILIPS BEDRIJVEN BV
Netherlands
Address
Philips Research Laboratories, Building Wb507
5600 MB Eindhoven
Radiotelevisione Italiana
Italy
Address
Corso Giambone 68
10135 Torino
SWISS TELECOM PTT
Switzerland
Address
Ostermundigenstrasse 93
3000 Bern 29
University of Essex
United Kingdom
Address
Wivenhoe Park
CO4 3SQ Colchester