Dynamic codebook adaptive vector quantization for image coding

Abstract Contemporary and future high spectral resolution sounders represent a significant technical advancement for environmental and meteorological prediction and monitoring. Given their large volume of spectral observations, the use of robust data compression techniques will be beneficial to data transmission and storage. In this paper, a novel adaptive vector quantization (VQ)-based linear prediction (AVQLP) method for lossless compression of high spectral resolution sounder data is proposed. The AVQLP method optimally adjusts the quantization codebook sizes to yield the maximum compression on prediction residuals and side information. The method outperforms the state-of-the-art compression methods [Joint Photographic Experts Group (JPEG)-LS, JPEG2000 Parts 1 and 2, Consultative Committee for Space Data Systems (CCSDS) Image Data Compression (IDC) 5/3, Context-Based Adaptive Lossless Image Coding (CALIC), and 3D Set Partitioning in Hierarchical Trees (SPIHT)] and achieves a new high in lossless compression for the standard test set of 10 NASA Atmospheric Infrared Sounder (AIRS) granules. It also compares favorably in terms of computational efficiency and compression gain to recently reported adaptive clustering methods for lossless compression of high spectral resolution data. Given its superior compression performance, the AVQLP method is well suited to ground operation of high spectral resolution satellite data compression for rebroadcast and archiving purposes.

Download Full-text

Adaptive vector quantization with fuzzy distortion measure for image coding

10.1117/12.238494 ◽

1996 ◽

Cited By ~ 3

Author(s):

Suryalakshmi Pemmaraju ◽

Sunanda Mitra ◽

L. Rodney Long ◽

George R. Thoma ◽

Yao-Yang Shieh ◽

...

Keyword(s):

Vector Quantization ◽

Image Coding ◽

Adaptive Vector Quantization ◽

Distortion Measure

Download Full-text

Adaptive Vector Quantization with Creation and Reduction Grounded in the Equinumber Principle

Journal of Advanced Computational Intelligence and Intelligent Informatics ◽

10.20965/jaciii.2005.p0599 ◽

2005 ◽

Vol 9 (6) ◽

pp. 599-606 ◽

Cited By ~ 6

Author(s):

Michiharu Maeda ◽

◽

Noritaka Shigei ◽

Hiromi Miyajima ◽

Keyword(s):

Neural Networks ◽

Vector Quantization ◽

Learning Process ◽

Image Coding ◽

Image Data ◽

Experimental Results ◽

Structural Learning ◽

Adaptive Vector Quantization ◽

Partition Space ◽

Predetermined Number

This paper concerns the constitution of unit structures in neural networks for adaptive vector quantization. Partition errors are mutually equivalent when the number of inputs in a partition space is mutually equal, and average distortion is asymptotically minimized. This is termed the equinumber principle, in which two types of adaptive vector quantization are presented to avoid the initial dependence of reference vectors. Conventional techniques, such as structural learning with forgetting, have the same number of output units from start to finish. Our approach explicitly changes the number of output units to reach a predetermined number without neighboring relations equalling the numbers of inputs in a partition space. First, output units are sequentially created based on the equinumber principle in the learning process. Second, output units are sequentially deleted to reach a prespecified number. Experimental results demonstrate the effectiveness of these techniques in average distortion. These approaches are applied to image data and their feasibility was confirmed in image coding.

Download Full-text