Color Image Steganography Scheme Based on Convolutional Neural Network

Image steganography has always an important part in the field of information security. In this paper, we propose a new color image steganography scheme base on convolutional neural network, which can hide the secret image into the cover image of the same size. The steganography scheme consists of two parts, the hiding network and the reveal network. Hiding network uses the skip connection, the low-level features of the image can be transmitted to each subsequent layer, which helps to hide the details of the secret image. Experimental results show that this scheme can not only achieve the steganography of color images, but also maintain a high embedding ability. Compared with other methods, the PSNR and SSIM mean values of our steganography image reach 43.6 dB and 0.987, the secret images extracted through the extracting network have also achieved good results.