Paper Abstract and Keywords |
Presentation |
2007-03-19 11:00
Near-Lossless Coding of Still Images Using Block-Adaptive Prediction Kazuya Takahashi, Nau Ozaki, Ichiro Matsuda, Susumu Itoh (Tokyo Univ. of Science), Sei Naito, Atsushi Koike (KDDI R&D Labs.) |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
This paper proposes a near-lossless coding scheme using block-adaptive prediction for still images. The scheme employs a quadtree-based variable block-size adaptive prediction technique to exploit spatial correlations in image signals which have various edges and textures. In order to keep magnitude of differences between the original and reconstructed images within a given limit of $d$, prediction errors obtained by the above prediction technique are quantized with a step-size of $2d+1$ and then encoded using a context-adaptive arithmetic coder. Simulation results indicate that the proposed method attains 8--50\,\% lower coding rates than the JPEG-LS standard when $d=1$. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Image coding / Near-lossless coding / Block-adaptive prediction / Linear quantization / Arithmetic code / / / |
Reference Info. |
IEICE Tech. Rep., vol. 106, no. 608, IE2006-278, pp. 17-22, March 2007. |
Paper # |
IE2006-278 |
Date of Issue |
2007-03-12 (IE) |
ISSN |
Print edition: ISSN 0913-5685 |
Download PDF |
|