A Novel Image Steganographic Framework for Efficient Embedding Capacity and Stego Quality

Abstract

Internet has emerged a major platform to support many activities in our daily life. However, the countless facilities offered by Internet brings numerous challenges thereof. One of the major challenge is regarding security and privacy of data transmitted over internet. Information security is an emerging area of research, in which steganography is a promising technique to hide the presence of secret information into the cover media such as, image, video, audio files. Although various steganographic schemes have been designed and proposed by researchers, however, the tradeoff among important steganography parameters, i.e., embedding capacity and Peak Signal to Noise Ratio (PSNR) remains challenging, because efforts to enhance the embedding capacity significantly degrades the PSNR and vice versa. Since, both these parameters have an important impact over the steganographic scheme therefore, a compromise on anyone may jeopardize the desired objectives. Hence this research fill-up this gap by proposing a novel steganographic framework that ensures an optimal tradeoff between embedding capacity and PSNR by achieving higher embedding capacity with lower visual quality distortions (PSNR), while comparing with current state of the art approaches. The proposed scheme uses Pixel Value Difference (PVD), Least Significant bits (LSB) and Difference Expansion (DE) techniques simultaneously and hide the secret data in the cover image with due emphasis on achieving higher embedding capacity, as well as, visual quality. The validity of proposed framework has been demonstrated using benchmark case study by choosing test images from publicly available Signal and Image Processing Institute (SIPI) dataset. The achieved results prove that the proposed scheme, enhances the embedding capacity by 1053858 bits (min) to 1061248 bits (max) (on average 0.27% to 0.63% more than the reported so far), while maintaining the visual quality within 32dB (min) - 36dB (max) (on average 1.69 to 5.21 dB improvement). Moreover, because of higher PSNR, our proposed scheme is more effective against histograms and statistical attacks due to improved visual quality of the stego images