Investigation on reversibility and quality of data hiding algorithms and its covert communication applications

Abstract

Digital Images form a major share of all the data communicated over the internet. With the advent of faster computational devices and high network bandwidth, securing digital data from unauthorized access is a demanding and resource intensive challenge. Traditional cryptographic algorithms can conceal data but cannot conceal covert communication. This paved the way for increased research in digital image data hiding technologies. The quality of data hiding algorithms is measured in terms of data embedding capacity, image quality, security features, reversibility and robustness of hidden data. High embedding can be achieved by using a combination of integer wavelet transform and modified prediction errorbased histogram shifting algorithm. The algorithm shifts the histogram of pixel prediction errors of images for embedding of data. The proposed sequential encoding based reversible data hiding algorithm uses sequential encoding for added security of hidden data by concealing the length of data. In the wavelet domain of image, conversion and round off errors are avoided by using appropriate lifting scheme so that coefficients are always integers. This made lossless reversibility of cover image possible. The provision for encryption of cover images enhances general usability of the algorithm. The wavelet-based image processing naturally allows for treating digital images as multiple sub-bands. This allows encrypting the image on scaled wavelet coefficient and providing provision for embedding of data in other sub-bands. This approach effectively allows encryption and data embedding to be performed in any sequence; independent of each other. The proposed algorithm can be fortified with BCH (63,7,15) error correction algorithm and pixel density based median pixel restoration for improving noise tolerance. Experimental studies show good noise mitigation performance. Finally, a system architecture is proposed that integrates all the developed and modified algorithms of this proposed work that are in alignment with the objectives of reversible data hiding in images. The algorithms are evaluated based on standard metrics like Peak Signal to Noise Ratio (PSNR), Mean Squared Error (MSE) and Structural Similarity Index (SSIM) and their performance characteristics are evaluated and verified newline