Summary
International Symposium on Nonlinear Theory and its Applications
2017
Session Number:B2L-B
Session:
Number:B2L-B-2
Scalable Lossless Image Coding Method Using Cellular Neural Networks with Greedy Template Optimization for Minimum Rate Coding
Hideharu Toda, Hisashi Aomori, Tsuyoshi Otake, Ichiro Matsuda, Susumu Itoh,
pp.481-484
Publication Date:2017/12/4
Online ISSN:2188-5079
Summary:
Because of development of digital archive and diversification of playback terminal, scalable lossless image coding techniques have become indispensable in resent years. However, in general, coding efficiency of scalable coding is inferior to that of a non-scalable framework. Therefore improvement of coding efficiency of scalable image coding method is very important task. In this paper, to deal with this difficulty, a hierarchical scalable image coding framework with cellular neural network (CNN) predictors that are trained to achieve high coding efficiency is proposed. In the proposed method, a CNN predictor having 12 parameters of each layer is optimized by greedy algorithm utilizing image pyramid. The proposed method was tested on various standard images for evaluating lossless coding efficiency. Experimental results support that coding efficiency of our method significantly outperforms that of conventional methods including the JPEG standards and latest techniques.