Skip navigation

Improved Registration and Segmentation for Image Analysis

Event Date: May 4, 2009
Speaker: Mark Pickering
Speaker Affiliation: School of Information Technology
and Electrical Engineering,
The University of New South Wales at the Australian Defence Force Academy,
Canberra, Australia
Sponsor: CNSIP Seminar
Time: 3:00 PM
Location: EE 118
Contact Name: Prof Edward Delp
Contact Phone: (765) 494-1740
Contact Email:





To retrieve higher meaning from images there are often a number of preliminary tasks that need to be performed.  Image registration and segmentation are two important processes that fall into this category.  In this seminar I will present a new approach for registration of a 3D object with a 2D image of the same or a similar object and a new approach for segmentation of images based on texture.  The new 2D-3D registration algorithm uses a similarity measure that is inherently robust to variations in intensity between the 2D and 3D objects and therefore can be used to register images of objects captured using different sensors.  The new texture segmentation algorithm is based on features derived from the dual-tree complex wavelet transform of the image and is able to perform segmentation that is invariant to the scale and rotation of the texture in the image regions.






Mark Pickering is an Associate Professor in the School of Information Technology and Electrical Engineering, The University of New South Wales, at the Australian Defence Force Academy.  Mark holds Bachelor, Masters and Doctor of Philosophy degrees in Electrical Engineering and has co-authored over 50 journal and international conference papers on image processing, compression and communications.  His main area of research is digital image processing.  Within this broad discipline, he has supervised research projects on more specific topics including: digital video compression, hyperspectral data compression, image segmentation, streaming video over computer networks, medical image registration, digital video watermarking and automatic target detection.