EUSIPCO'2002 banner

Paper data
-----
Title:
Progressive coding in JPEG2000 - Improving content recognition performance using ROIs and importance maps

Author(s):
Nguyen Anthony, Queensland University of Technology, Australia
Chandran Vinod, Queensland University of Technology, Australia
Sridharan Sridha, Queensland University of Technology, Australia
Prandolini Robert, Defence Science and Technology Organisation, Australia

Page numbers in the proceedings:
Volume III pp 197-200

Session:
JPEG 2000 (2/2)

Paper abstract
-----
This paper presents a system, which permits arbitrary regions of importance and corresponding degrees of importance to be assigned in an image within the JPEG2000 image coder. The image can be reconstructed progressively in layers prioritised by importance values rather than the objective criterion of overall PSNR optimisation. Importance progressive coding is very desirable in surveillance applications. The system, IMP-J2K, aims to improve the interpretability or content recognition performance of compressed images, by selecting and distributing the bit allocation to contents of importance in an image and embedding these earlier in the code-stream. IMP-J2K makes use of JPEG2000s data packing sub-system, which can selectively encode data in layers.

Paper
-----
A PDF version is available here

-----