An image compression method based on multiple models for the probabilities of patterns

✍ Scribed by Yung-Kuan Chan; Ching-Lin Wang

• Publisher: John Wiley and Sons
• Year: 2009
• Tongue: English
• Weight: 456 KB
• Volume: 19
• Category: Article
• ISSN: 0899-9457
• DOI: 10.1002/ima.20214

✦ Synopsis

Abstract

This article proposes an image compression method based on multiple models for the probabilities of patterns (MMPP method) to encode a gray‐level image f. First, the MMPP method employs a median edge detector (MED) to reduce the entropy of f. The intensities of two adjacent pixels in an image are usually close to each other. A base switching transformation (BST) is then used to lessen the spatial redundancy of f. Finally, the arithmetic encoding method is applied to further encode the data generated after the processing of MED and BST. To reduce the memory space required to hold f, the MMPP method classifies the data and then compresses the data in each cluster by the arithmetic encoding method based on different probability tables. The experimental results show that mostly the MMPP method can provide better efficiency in memory space than the lossless JPEG 2000 method does. © 2009 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 19, 362–368, 2009