Optimal image compression via block-based adaptive colour reduction with minimal contour effect |
| |
Authors: | Lüsi Iiris Bolotnikova Anastasia Daneshmand Morteza Ozcinar Cagri Anbarjafari Gholamreza |
| |
Affiliation: | 1.iCV Research Group, Institute of Technology, University of Tartu, Tartu, 50411, Estonia ;2.School of Computer Science and Statistics, Trinity College Dublin, Dublin 2, Ireland ;3.Department of Electrical and Electronic Engineering, Hasan Kalyoncu University, Gaziantep, Turkey ; |
| |
Abstract: | Current image acquisition devices require tremendous amounts of storage for saving the data returned. This paper overcomes the latter drawback through proposing a colour reduction technique which first subdivides the image into patches, and then makes use of fuzzy c-means and fuzzy-logic-based inference systems, in order to cluster and reduce the number of the unique colours present in each patch, iteratively. The colours available in each patch are quantised, and the emergence of false edges is checked for, by means of the Sobel edge detection algorithm, so as to minimise the contour effect. At the compression stage, a methodology taking advantage of block-based singular value decomposition and wavelet difference reduction is adopted. Considering 35000 sample images from various databases, the proposed method outperforms centre cut, moment-preserving threshold, inter-colour correlation, generic K-means and quantisation by dimensionality reduction. |
| |
Keywords: | |
本文献已被 SpringerLink 等数据库收录! |
|