A novel source modeling framework for bit rate and picture quality control in DCT visual coding

Ahihai He, Yong Kwan Kim, and Sanjit K. Mitra
Dept. of Electrical and Computer Engineering
University of California, Santa Barbara
{zhihai, ykkim, mitra} [at] iplab.ece.ucsb.edu

Abstract

We present a novel source modeling framework in this paper. With the proposed source model, we can estimate the rate-distortion(R-D) curve before quantization and coding for DCT-based still images and video compression. The relative estimation error is less than 5%. In addition, the estimation algorithm has very low computational complexity. Based on the estimated R-D curve, we can perform accurate bit rate and picture quality control for still images and video compression.
[PDF] [BibTex]
Zhihai He, Yong Kwan Kim, and Sanjit K. Mitra,
Picture Coding Symposium, South Korea, Apr. 2001.
Node ID: 320 , DB ID: 117 , Lab: IPL , Target: Conference
Subject: [IPL] « Look up more