Wavelet Domain Digital Image Hiding Algorithm Based On CNN Encryption

As the network is becoming more and more popular today, the exchange of information has got to an unprecedented depth and breadth. The sensitive information involving military, political and commercial confidentiality is also transmitted through the public network, and it also needs protection. Then the information hiding technology is produced to serve the concealment communication and the protection of property rights.The information hiding technolgy is mainly used in the image hiding domain for the three advantages, which are intuitive imgae, large amount of information and convenience for caculation. The current image hiding technique is mainly made up of the preprocessing algorithm, embedding algorithm and extraction. The preprocessing algorithm is used to encrypt the original secret image in order to improve the security of the entire system. The traditional chaotic encryption system has many disadvantages, like relatively simple dynamic characteristics, few of parameters, simple keys and low security, which determine the confidentiality of the chaotic sequence is not enough.The fifth-order cellular neural network (CNN) is used to produce highly sensitive and complicated hyperchaotic sequences, which are quantizated and coded into binary hyperchaotic sequences, then make up the optimized CNN encryption algorithm with the arnold encryption algorithm. Based on the similarity of the low bit plane and random noise, a discrete wavelet transforms (DWT) domain embeding algorithm has been proposed which is in view of the bit plane compensation method. Embed four high bit planes which have been treated by the CNN encryption, when the receiver extracts the secret information, four secret high bit planes are extracted firstly, then choose a gray image randomly with the same size as the secret image, extract the four low bit planes of the gray image, and recover the original secret image combining with the four original secret high bit planes. The experiment results show that the invisibility is highly improved by reducing the hiding capacity and achieve good robustness at the same time.