Adaptive PVD-MPK encoding method in transform domain with a modified side match method against dynamic and constant attacks

• Published in: Multimedia tools and applications 2022-07, Vol.81 (16), p.23493-23532
• Main Authors: El Salam, Nada Hussien Abd, Ismael, Ahmed Hamdy, Xiong, Shengwu
• Format: Article
• Language: English
• Subjects: Coefficients; Computer Communication Networks; Computer Science; Data Structures and Information Theory; Discrete Wavelet Transform; Domains; Embedding; Entropy (Information theory); Image analysis; Image quality; Multimedia Information Systems; Pixels; Special Purpose and Application-Based Systems; Steganography; Wavelet transforms
• ISSN: 1380-7501; 1573-7721
• DOI: 10.1007/s11042-022-11995-0

This paper presents a novel image steganography technique that uses a mixture of pixel value differences and mobile phone keyword (PVD-MPK) encoding based on adaptive range tables and a modified side match method to conceal a secret message in either the spatial domain or an integer-wavelet-transform (IWT) based transform domain of the cover image. First, the method uses IWT instead of discrete wavelet transform (DWT) to achieve accurate extraction of secret bits. Then, the low-low (LL) sub-band is divided into nonoverlapping 2 × 2 coefficient blocks that are used as an embedding region. Second, the first coefficient of each block is paired with surrounding coefficients in three directions. Then, the diagonal and vertical directions, respectively, are selected for embedding a secret message using the PVD-MPK method and for embedding a secret message using a modified side match method if the difference between the new coefficient pixels in the diagonal direction is equal to or greater than a threshold (predefined by users or automatically determined by image analysis) to check whether or not this block is an edge or sharpest block. Experimental results show that the performance of the proposed method is superior to that of existing methods in terms of both data hiding capacity and image quality. The proposed method can reach a maximum of 1,223,949 bits with a PSNR value of 44.12 dB in the spatial domain and 337,304 bits with a PSNR value of 42.48 dB in the transform domain. Furthermore, NPCR, UACI, information entropy, and CC values were 99.62, 28.16, 7.997, and − 0.0041, respectively, so it maintains a good level of security and is challenging to be detected with human vision.