Optimal reconstruction value for DCT dequantization using Laplacian pdf model

So-Yeon Kang; Byung-Uk Lee

doi:10.1117/12.526684

18 January 2004 Optimal reconstruction value for DCT dequantization using Laplacian pdf model

So-Yeon Kang, Byung-Uk Lee

Proceedings Volume 5308, Visual Communications and Image Processing 2004; (2004) https://doi.org/10.1117/12.526684
Event: Electronic Imaging 2004, 2004, San Jose, California, United States

Abstract

Many image compression standards such as JPEG, MPEG or H.263 are based on the discrete cosine transform (DCT), quantization, and Huffman coding. Quantization error is the major source of image quality degradation. The current dequantization method assumes the uniform distribution of DCT coefficients. Therefore the reconstruction value is the center of each quantization interval. However DCT coefficients are regarded to follow Laplacian probability density function (pdf). We derive an optimal reconstruction value in closed form assuming Laplacian pdf, and show the effect of the correction on image quality. We estimate the Laplacian pdf parameter for each DCT coefficient, and obtain a correction for reconstruction value from the proposed theoretical predictions. The corrected value depends on the Laplacian pdf parameter and the quantization step size Q. The effect of PSNR improvement due to the change in dequantization value is about 0.2 ~ 0.4 dB. We also analyze the reason for the limited improvements.

Citation Download Citation

So-Yeon Kang and Byung-Uk Lee "Optimal reconstruction value for DCT dequantization using Laplacian pdf model", Proc. SPIE 5308, Visual Communications and Image Processing 2004, (18 January 2004); https://doi.org/10.1117/12.526684

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available