A Study of Coding Parameters and Objective Picture Quality Scale for High Definition Still Pictures Coded by JPEG Scheme

Abstract

This paper describes the relationship between picture quality in high definition still pictures and a set of coding parameters in the JPEG scheme. The Huffman code table and the quantization matrix is adjusted for high definition still pictures. This adjustment leads to a great improvement in picture quality, especially at a high compression ratio. The quality of a picture coded with the same coding parameters varies significantly according to the picture content. If automatic picture quality assessment and automatic selection of coding parameters suitable for each picture is enabled, program production will become very efficient.
An objective picture quality scale for high definition still pictures is developed. The scale is derived from the following three distortion factors, 1) weighted noise, 2) block distortion, and 3) mosquito noise. The objective picture quality scale and the quality of the subjective pictures obtained experimentally show good agreement.