18 April 2006 Derivate statistics for Karhunen-Loeve transform
Author Affiliations +
Proceedings Volume 6180, Photonics, Devices, and Systems III; 618020 (2006) https://doi.org/10.1117/12.675849
Event: Photonics, Devices, and Systems III, 2005, Prague, Czech Republic
Abstract
This paper introduces video compression and image processing methods for fast multimedia channels. Standards designed for image and video coding (JPEG2000, MPEG-4 and MPEG-7), being not strictly defined, present a large field for further experiments and improvements. Our method was developed on fundamentals of Vector Quantization (VQ) and the most advantageous qualities of Karhunen-Loeve Transform (KLT) (known also as Principal Component Analysis - PCA). We also present a direct an idea of how to implement optical transform that can be used with accordance to the methods presented. Among other compression techniques, KLT optimality is discussed as a purpose for further improvement to the method.
© (2006) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Pavel Hanzlík, Petr Páta, "Derivate statistics for Karhunen-Loeve transform", Proc. SPIE 6180, Photonics, Devices, and Systems III, 618020 (18 April 2006); doi: 10.1117/12.675849; https://doi.org/10.1117/12.675849
PROCEEDINGS
7 PAGES


SHARE
RELATED CONTENT

4K-based intra and interprediction techniques for HEVC
Proceedings of SPIE (April 29 2016)
Seam carving for semantic video coding
Proceedings of SPIE (September 23 2011)
Stable buffer control strategy with learning capability
Proceedings of SPIE (April 21 1995)
Future trends in image coding
Proceedings of SPIE (January 12 1993)
Accelerating M-JPEG compression with temporal information
Proceedings of SPIE (December 29 1997)

Back to Top