Research On Hybrid Image Encryption Algorithms Based On Compressive Sensing

Recently,with the rapid development of network technology and multimedia technology,the security problem in the process of image transmission and storage stands out.Unlike text,an image itself has big data,strong correlation among adjacent pixels and unequal energy distribution.In the case of limited bandwidth /storage space,we have to consider compression as well as encryption while transmitting/storing large images.However,traditional encryption algorithms cannot achieve the process of compression and encryption simutaneously.In addition,for image encryption,traditional encryption algorithms often have low efficiency and cannot remove the correlation between adjacent pixels.Therefore,the exploitation of encryption algorithm tailored for images has become very necessary.As compressive sensing can achieve high compression ratio as well as encryption,it has been widely applied in the field of image encryption since it was proposed.But simple image encryption algorithms only based on compressive sensing usually have large amount of key transmission and are vulnerable to security attacks.At the same time,optics and chaos are the two most important and classical techniques for image encryption due to their essential features.To improve encryption performance,domestic and overseas scholars have proposed various hybrid image encryption algorithms based on the combination of compressive sensing,chaos and optics according to their characteristics.Simulations and theoretical analyses verify that the hybrid image encryption algorithms is superior to these based on single technology in performance.In this dissertation,the hybrid image encryption schemes based on compressive sensing,chaos and optics are summarized and classified into six basic frameworks.The advantages,disadvantages and the improving directions of the six image encryption frameworks are analyzed and discussed.What's more,on the basis of the latest hybrid encryption framework,we propose a self-adaptive encryption scheme for double color images based on compressive sensing,discrete fractional random transform and one-dimensional compound chaotic maps.Initially,two plaintext images are measured and encrypted by compressive sensing based on one-dimensional compound chaotic maps.Then we combine the two measured real images into one complex image.Finally,the complex image is reencrypted by double random phase encoding to obtain final cipher matrices based on one-dimensional compound chaotic maps and discrete fractional random transform.Considering the process of compressive sensing and most of the double random phase encoding are linear transformations,they usually fail to resist common attacks.To solve the problem,we introduce self-adaption into the process of double random phase encoding.In the process of double random phase encoding,RGB color components interact with each other and the generation of key sequences are not only controlled by chaotic maps,but also by the plaintext images.Simulations and theoretical analyses show that this algorithm has high compression ratio and high safety level.