Research On High-capacity Reversible Data Hiding Algorithm Of JPEG Image In Encryption Domain

With the development of cloud storage and network communication technology,massive image data began to be transmitted on the public network,but th e authenticity and integrity of multimedia data is difficult to be guaranteed.Reversible data hiding in encrypted images has the advantages of both image encryption technology and reversible data hiding technology,which greatly protects the privacy of mu ltimedia data.For some special application scenarios,such as medical treatment,law,military and other scenarios,it is necessary to hide the secret data in the image carrier in an imperceptible way for transmission,and the cover image and secret data can be recovered at the Receiving terminal without any sequence or damage.Therefore,it is necessary to combine image encryption technology with reversible data hiding technology to realize secret communication and improve the security of transmitted image data.Aiming at the problems of low embedding capacity and weak robustness of existing reversible data hiding algorithms for JPEG ciphertext images,this thesis studies the key technologies such as high efficiency,high fidelity,robustness,high capacity and security by using chaotic encryption,quadtree and 2-D discrete cosine transform(2D-DCT).The specific research work is as follows:1.In order to improve the low efficiency and poor robustness of the existing JPEG image encryption algorithm,as well as improve the anti-attack ability and key space of the encryption algorithm,JPEG image encryption algorithm based on hyperchaos,mixed hash and dynamic DNA is proposed.The plaintext image is converted into a DNA matrix by using DNA coding and decoding rules,and then the encryption key is replaced by the mixed hash function composed of MD5 and SHA-256 and the encryption key calculated by chaotic system.Then the DNA operation is dynamically controlled by the encryption key to carry out the diffusion op eration.Finally,the DNA encoding and decoding rules are used to obtain the ciphertext image.Experimental results show that the algorithm has good security and encryption efficiency,and the key space can reach 2⁵⁰⁷,which can effectively resist all kind s of attacks.At the same time,the key sensitivity is high,the pixel correlation coefficient is close to 0,the information entropy is close to 8,the UACI and NPCR values are close to ideal values,and it has good robustness to noise attack,cropping attack and JPEG compression attack.2.In order to solve the problems of the existing reversible data hiding methods of JPEG ciphertext images,such as low embedding capacity,poor robustness and poor visual quality of image recovery under noise attack or cropping attack,a reversible data hiding algorithm of high capacity JPEG ciphertext images based on 2D discrete cosine transform(2D-DCT)and quadtree segmentation was proposed.The algorithm preprocesses the original cover image by using the quadtree segmentation principle.Then the encryption key is replaced by the mixed hash function composed of MD5 and SHA-256 and the encryption key calculated by chaotic system.Finally,the embedded data is encoded using Huffman coding to obtain the data to be embedded,and the 2D-DCT transform and inverse transformation are combined to generate the masked secret text image.At the Receiving terminal,the receiver can extract the secret data and restore the cover image without any damage.Experimental results show that the algorithm achieves higher embedding capacity and stronger robustness.When the embedding capacity is 2.653bpp,the average peak signal-to-noise ratio of the test images reaches more than 41dB.