Current page: Members->List of Members->Edwin Hancock
 

Edwin Hancock
University of York
Membership Number: 40
Address: United Kingdom
Email: erh@cs.york.ac.uk
Phone: (44)
Fax: (44)
URL: http://www.cs.york.ac.uk/~erh/cvprgroup/index.html

Biographical Sketch
Professor Edwin Hancock (B.Sc in Physics, 1977, PhD in High Energy Physics, 1981, from the University of Durham) has been with the Department of Computer Science at the University of York since 1991, and has been Professor of Computer Vision since 1998. He has 17 years of experience in the fields of computer vision, image analysis and pattern recognition. His research has focussed on the use of graph-representations in high and intermediate level vision, and in the development of probabilistic methods for solving constraint satisfaction problems. He has published some 350 scientific papers, including 75 in international journals. He serves on the editorial boards of the journals IEEE Transactions on Pattern Analysis and Machine Intelligence, and Pattern Recognition. He has chaired six conferences and workshops, and has served on the programme committees of a number of international conferences.He has won best paper awards from the Pattern Recognition journal (twice), ACCV and CAIP. He is a fellow of the IAPR.

University of York (Computer Vision Group)
The Computer Vision and Pattern Recognition Group at the University of York focuses on shape analysis and recognition. Current research concentrates on the use of structural representations in high level vision, shape-from-shading and texture, surface representation and analysis, visual learning, and, object recognition. The methodologies used in this work are drawn from probability theory and optimisation. An important theme running through this work has been to develop probabilistic methods for describing structural variations in graph-representations of 2D image structure. This framework has been used to develop a statistically robust algorithms for correspondence matching and object recognition. It has been used in conjunction with a number of optimisation methods including genetic search, tabu search and mean-field annealing. More recently, our efforts have shifted to applying the framework to the problem of learning structural descriptions of shape. Here we have combined ideas from spectral graphtheory and probabilistic graphical models. The work is finding application in learning shape-classes from structural representations such as shock-trees and view graphs, and in retrieving images from large data-bases.

This work is complemented by a programme of activity which focuses on the recovery of 3D surface shape using shading and texture cues. At the low level end, we have developed new geometrically-based methods for estimating surface orientation and local surface topography using shading and texture gradients. Recent work has extended this work to accommodate complex reflectance and light scattering models. The shape-information delivered by these processes have been used for a number of high level vision tasks. These include view synthesis and interpolation, object recognition and model acquisition. In addition to these two main research strands, we have activity in the areas of perceptual grouping, content-based image retrieval, image watermarking, motion analysis and low-level feature detection. The group currently consists of three academics (Bors, Hancock and Wilson), three research associates and ten PhD students. It is based in a Department that received the top grade the last three UK Research Assessment Excercises. The groups activities commenced in 1991.It has been successful in obtaining 9 major EPSRC grants and about GBP 1M from industry and government. Members of the group are on the editorial boards of the journals IEEE TPAMI, IEEE TNN and Pattern Recognition.

A full description of the groups activities can be found at

http://www.cs.york.ac.uk/~erh/cvprgroup/index.html


Site generated on Friday, 06 January 2006