A pixel block fuzzy similarity measure applied in two applications

In this paper, the fuzzy set theory is used to represent the "vague" concept of a pixel block similarity within its spatial context. A similarity measure is proposed and used in two applications. The first one relies on this measure to conceal errors occurring at the decoding stage of a vector quantization (VQ) based image transmission system. The second application is a new visual encryption approach based upon the proposed measure. In this approach, only a part of the image is considered as confidential. The similarity measure is then used to substitute automatically this information by another one of innocent meaning while preserving the whole image coherence. The original information is embedded within the modified image, in a watermarking or steganographic way. Simulation results show that the proposed similarity concept considerably improves the subjective quality of VQ images transmitted over a noisy channel. It is also shown that it is possible to replace automatically a region of an image.