Variable-rate transform coding that controls quality of image using human visual model

✍ Scribed by Haruyuki Harada

• Book ID: 104591594
• Publisher: John Wiley and Sons
• Year: 1993
• Tongue: English
• Weight: 725 KB
• Volume: 24
• Category: Article
• ISSN: 0882-1666
• DOI: 10.1002/scj.4690240710

✦ Synopsis

Abstract

This paper defines a quantitative measurement SNRq for the evaluation of image quality in the discrete cosine transform domain considering human visual characteristics. A new coding algorithm which minimizes the bit‐rate for a given target SNRq also is obtained. When coding the image using the proposed algorithm, the bit‐rate varies according to the image busyness, but this does not present a problem because the image is stored in memory. Furthermore, it is not a problem when transferring the image if only the packet‐switched networks are used. Using computer simulation, the value of this quantitative measurement (the target SNRq) is found so that the differences between the original and the processed images were negligible.