Literature Report

Note: Figures were excluded from this HTML compilation. The version which contains all figures is a PDF

Abstract:

Construction of robust and concise statistical models is a goal which is difficult to achieve due to the high computational requirements it typically imposes. Numerous attempts have been made to automate the construction of good active appearance models, but none has yet been successful. Potential is born, however, in the unification of model construction and image registration. The evaluation of non-rigid registration, which is based on non-linear warps, has been another issue of great difficulties and automatic selection of good warps is far from trivial.

In active appearance models the main problem is the inability to select good landmarks without human judgement, as well as the difficulty in location and annotation of these landmarks using brute-force only. Non-rigid registration is a quickly emerging technique that can be used to warp multiple images and give some group-wise optimal model, as opposed to a model derived from pair-wise registration that depends on an arbitrary choice of a reference image.

These arguments highlight the benefits summoned by the combination of these two techniques - active appearance model can aid the selection of good warps in non-rigid registration and the functionality of non-rigid registration can help obtain more compact and robust models of appearance. This report outlines the current work in the field as well as some concepts that bear potential or whose realisation can contribute to future endeavours. It critically surveys some existing work and techniques along with the results they have produced. Lastly, this report attempts to identify some existing gaps where substantial improvements can be still made and describe the research planned with its aims and intermediate milestones.

Literature Report

R. S. Schestowitz
Imaging Science and Biomedical Engineering
Stopford Building
University of Manchester
United Kingdom