Image authentication using generalised finite automata

Abstract

Listen

Translate article

An effective technique for combined fragile and semi-fragile image authentication is presented. The authentication signature is extracted from image representation in the automata domain, unlike most conventional methods that are based on frequency domain. In the automata domain, the image is represented in terms of its perceptual self-similarity and can therefore survive high JPEG compression ratios. This method is also capable of changing the degree of fragility by setting the value of the image self-similarity parameter α. Furthermore, the proposed technique features an extremely compact signature length while maintaining a rather simple implementation. Experimental results are presented to demonstrate the effectiveness of the proposed technique.