Efficient transform-domain size and resolution reduction of images

摘要

This paper describes techniques for adapting a transform-based image (such as JPEG) so that it conforms to specific compressed size or dimension criteria. For compressed size adaptation, the proposed method operates efficiently in the transform domain, but yields visually improved results when compared to existing approaches. For resolution adaptation, the proposed method results in a speed improvement of 76% over the baseline algorithm and 31% over other previous work, with minimal or no difference in overall visual quality.