RGB Based Secure Share Creation in Steganography with ECC and DNN

Abstract

AbstractImage steganography is a technique for concealing sensitive data inside an image. Since traditional image steganography encrypts sensitive data in the host image, the payload capability is often disregarded, and steganographic image quality for the Human Visual System (HVS) must be improved. As a result, we proposed a new deep learning-based multiple image steganography approach in this work. The secret image is embedded into the cover image using a Deep Neural Network (DNN), and the embedded cover image is then encrypted using Elliptic Curve Cryptography (ECC) is used to improve the image’s anti-detection properties. The Deep Neural Network is a combination of Pre-processing, Hiding, and Revealing networks that enable steganography and extraction of full-size images to enhance steganographic performance. The results of the experiments show that the network can successfully embed up to three secret images into the cover image without causing a major difference in the Structural Similarity Index (SSIM). Furthermore, the image created using this steganography method has higher SSIM and Peak Signal to Noise Ratio (PSNR) values, with values of 0.99 and 35 dB, respectively.KeywordsImage steganographyElliptic Curve Cryptography (ECC)Deep Neural Network (DNN)Convolutional Neural Network (CNN)Information securityDigital data security