Extended Adaptive Pixel Pair Matching for data hiding in medical images

Abstract

Abstract

Data embedding in medical images require perfect reversibility with which the embedding distortions can be completely removed to achieve a high fidelity recovery of the cover images. Adaptive Pixel Pair Matching is a high capacity embedding scheme that allows data embedding in any desired larger notational system still achieving exact reconstruction of secret. The principle behind Extended Adaptive Pixel Pair Matching (EAPPM) is to use a pixel pair as the reference co-ordinate and select a co-ordinate in its neighbourhood set according to the secret digit, as the stego pixel pair using which the extraction function recovers the secret data. The cover image is then reconstructed using the Look-Up Table (LUT) built with values of modulus distance. This reversibility can lead to multiple layer embedding, but cannot be fully exploited to increase the payloads as the usage of larger notational systems reduces the quality of the image. Hence this paper extends the APPM process to be performed over coefficients of the Transform domain in a repetitive fashion improving embedding capacity still maintaining high image quality as required for medical images.