Research On Hiding Images In Images Based On Residual Attention Networ

Nowadays,the data that people generate on the Internet is doubling every 18 months,and the largest proportion of this data is in the form of images.Under the background of the huge influx of network resources,people are also faced with the huge challenge of information security.Information security-related research with images as the main data carrier has received the attention of a large number of researchers,and one of the important areas is to hide pictures with pictures.Hiding images into images technology is a covert communication technology that disperses,embeds and fuses large-capacity information that needs to be hidden into another carrier image.The development of this covert communication technology benefits from the rapid development of deep learning technology in recent years,especially Convolution Neural Networks(CNNs).Compared with the traditional information hiding method designed by manual distortion function,the image hiding image technology has the advantage of large hiding capacity.However,almost all current directly reconstruction based methods directly reconstruct dense images through autoencoder architectures.The dense images obtained by such methods are often of low visual quality,prone to fading,blurring and other phenomena,and it is difficult to accurately restore the hidden images.In this paper,through the in-depth study of the problem of hiding a map with a map,the essence of the task of hiding a map with a map is to solve a conditional optimization problem.There may be two reasons for the inability to obtain better dense images: first,the method of directly reconstructing dense images is to solve the optimization problem from a huge analytical domain;second,the hiding and extraction of information in the image is a process of interfering with each other.In order to solve the above-mentioned problems,this paper explores from the perspectives of the task of hiding images from images and deep learning technology,and proposes an improved method of hiding images from images in depth.