A Quadsection Algorithm for Grammar-Based Image Compression
-
- HAYASHIDA MORIHIRO
- Bioinformatics Center, Institute for Chemical Research, Kyoto University
-
- RUAN PEIYING
- Bioinformatics Center, Institute for Chemical Research, Kyoto University
-
- AKUTSU TATSUYA
- Bioinformatics Center, Institute for Chemical Research, Kyoto University
Search this article
Abstract
For the purpose of text compression, grammar-based compression, which is to find a small grammar that generates a given data, has been well-studied. In this technical report, we apply this methodology to compression of rectangular image data. We first define a context-free rectangular image grammar (CFRIG) by extending the context-free grammar. Then we propose a quadsection type algorithm by extending a bisection type algorithm for grammar-based compression of text data. We show that our proposed algorithm approximates in polynomial time the smallest CFRIG within a factor of O(n4/3), where an input image data is of size O(n) × O(n). We also present results on computational experiments on the proposed algorithm.For the purpose of text compression, grammar-based compression, which is to find a small grammar that generates a given data, has been well-studied. In this technical report, we apply this methodology to compression of rectangular image data. We first define a context-free rectangular image grammar (CFRIG) by extending the context-free grammar. Then we propose a quadsection type algorithm by extending a bisection type algorithm for grammar-based compression of text data. We show that our proposed algorithm approximates in polynomial time the smallest CFRIG within a factor of O(n4/3), where an input image data is of size O(n) × O(n). We also present results on computational experiments on the proposed algorithm.
Journal
-
- 情報処理学会研究報告. MPS, 数理モデル化と問題解決研究報告
-
情報処理学会研究報告. MPS, 数理モデル化と問題解決研究報告 78 S1-S6, 2010-05-21
- Tweet
Details 詳細情報について
-
- CRID
- 1570572702598115328
-
- NII Article ID
- 110007993903
-
- NII Book ID
- AN10505667
-
- ISSN
- 09196072
-
- Text Lang
- en
-
- Data Source
-
- CiNii Articles