Details of Grant 

EPSRC Reference:	GR/J79997/01
Title:	PARALLEL MOTION ESTIMATION TECHNIQUES FOR NOISY IMAGE SEQUENCES
Principal Investigator:	Martin, Dr G
Other Investigators:	Packwood, Dr R
Researcher Co-Investigators:
Project Partners:
Department:	Computer Science
Organisation:	University of Warwick
Scheme:	Standard Research (Pre-FEC)
Starts:	01 November 1994	Ends:	31 December 1996	Value (£):	81,166
EPSRC Research Topic Classifications:	Digital Signal Processing
EPSRC Industrial Sector Classifications:
Related Grants:
Panel History:
Summary on Grant Application Form
To improve the compression/fidelity performance of coding noisy time-varying image sequences using motion estimation techniques explicitly designed for parallel implementation.To achieve the above by refining existing optical flow techniques, and to use the vector information thus derived to construct non-uniform block structures of moving areas. To investigate appropriate parallel algorithms for both general purpose and application-specific implementations.Progress:The project is at a very early stage, but results of initial investigations clearly demonstrate the poor performance of conventional block-matching techniques for estimating motion in noisy image sequences. Under considerably noisy conditions the resulting motion vectors are almost random, thus decreasing the compression ratio and image quality. Therefore in order for the motion estimation to be robust under noise, the estimation algorithms have to rely on the true motion present in the image sequence rather than merely best-matching the data within blocks. There appears to be no doubt that optical flow estimation techniques can be applied to good effect. A new algorithm has been developed which, although based on block-matching, does not exclusively depend on a conventional matching criterion (e.g. Mean Absolute or Mean Square Error). Instead, a multiresolution approach is adopted, where the motion extracted from matching comparatively small blocks, representing high spatial resolution, is used to grow regions (variable-sized blocks) containing uniform motion. The process is analogous to optical flow, and the algorithm is robust in the presence of noise. It is also computationally economic as it does not involve any demanding operations such as relaxation. Application of the technique on real image sequences to which noise has been added has resulted in improvements in signal-to-noise ratio of up to 10% over conventional fixed block matching. The algorithm has performed well in a simple DCT codec, but more significantly as it successfully extracts true motion information it can be used in model-based hybrid coding schemes, of the type likely to be required for very low bit-rate systems.
Key Findings
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Potential use in non-academic contexts
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Impacts
Description This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk Summary Date Materialised
Sectors submitted by the Researcher
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Project URL:
Further Information:
Organisation Website:	http://www.warwick.ac.uk