Equivalent representations and generalizations of the image resolution conversions in DCT domain

✍ Scribed by Xiaoxia Zou; Hitoshi Kiya

• Publisher: John Wiley and Sons
• Year: 2000
• Tongue: English
• Weight: 729 KB
• Volume: 83
• Category: Article
• ISSN: 1042-0967
• DOI: 10.1002/(sici)1520-6440(200003)83:3<31::aid-ecjc4>3.0.co;2-j

✦ Synopsis

Image resolution conversions are methods for converting a digital image to another with a different number of pixels. In this paper, two image resolution conversion methods in the DCT region for image minification are discussed. The images minified by these methods are known to have different picture quality. Specifically, one of them has the advantage of excellence in realization while the picture quality is known to be poor. However, the cause of the differences in picture quality has not been found. In this paper, the equivalent representations of these conversion methods in the time domain are derived. Based on these equivalent representations, the cause of the difference in picture quality in the two methods is theoretically explained. Further, by extending the conventional method, a generalized image resolution conversion method in the DCT domain is proposed. As a result, it becomes possible to improve the picture quality in the conventional method.