Enhanced and secured hybrid steganography model for hiding large data

Abstract

Recent works on steganography basically are focused on various network layers and multiple data-hiding techniques. These researches lead to image manipula- tion, contortion and small-scale payload. This paper proposes a new dimension of the steganography model by merging the techniques of hiding texts and images. In- stead of hiding only one type of data this proposed model is focusing on veiling two

types of data (first text and then image) by masking one data (text) into another stego data (image) and then covering them both into a cover image. This model ensures a minimal peak-signal-to-noise ratio (PSNR) and no noise. This will be a model of suppressing more data in a stego container for one cover image using AES,

MLSB and LSB respectively. This also should increase the payload capacity of em- bedding images through the steganographic system architecture with recognition.

Here, we have used a pair of encoders and decoders for encryption and decryption to take two inputs and generate one hybrid output. The reverse of the encryption process will do the work for decrypting and decoding particular cryptic data. The AES (Advanced Encryption Standard) algorithm and LSB (Least Significant Bit) have been used to ensure our proposed model’s accuracy and effectiveness. More- over, using a modified 16-bit key is ensuring the safety of any confidential data of the

user. It also accomplishes better execution by using ImageNet datasets. This model will expand and escalate the safety mechanism of steganography for concealing a larger amount of hybrid data. Furthermore, it can decide how many bits need to be changed in LSB depending on the length of the text that has to be encrypted in the host image. It is a hierarchy of two steganographic and one encrypt model. Last but not least, it makes certain of a distortion-free process where retrieving data after successfully concealing it is fruitful depending on higher accuracy, SSIM, PSNR and lower MSE.