Research On Low Complexity Image Encryption Based On Compressed Sensing

With the development of Internet and Mobile Internet, the multimedia signalcommunication，especially image signal has gradually became a dominant form inmodern network communication. However, characteristics of image such as large dataamount and high redundancy always result in calculation amount problem. How totransmit image signals accurately and quickly has become a research hotspot incommunication engineering field. Moreover, Internet brings convenience to human aswell as urgent network security problems and communication reliability problems. Forexample, information stealing, interception, malicious modification, and packet loss, etc.Aiming at the above three questions, this paper gives a specific solution by introducinga new kind of technology called Compressed Sensing.Compressed Sensing (CS) is a new proposed technology which can process allsparse signals and compressible signals. The later ones cover most of the signals innature. It uses little linear projection to recover signals with a sampling frequency farbelow the traditional Nyquist sampling law defines, thus decreasing the sampling andcomputing cost of the transmitting terminal and improving process efficiency. So CS isvery suitable to be applied in the image communication system which is of large dataamount and high redundancy. As a result, CS based image communication system oflow complexity has become an important branch of CS applications. Moreover, themeasurement matrix of CS can be designed by a secret key to accomplish encryptionprocedure, as long as it satisfies relevant rule and RIP principle. So CS has property ofencryption, and can be used to design encryption system. Last but not least, CS basedencryption system can decrypt highly incomplete signals with high probability, even ifserious packet loss occurs. While in this circumstance, traditional encryption technologycannot do such well as CS. So CS encryption system has strong anti-packet lossreliability.According to the above reasons, the paper presents a low-complexity, yethigh-performance CS image encryption system. The core technology is namedseparate-combine decryption, which is proposed to couple the existing separablesensing method of the transmitting terminal. In the “separate” stage, the compressedrows(or columns) data will be recovered first to get an intermediate variable, whosecolumns(or rows) data will then be recovered to get the row(or column)-first-decryption-result. In the “combine” stage that follows, the final result will beobtained by weighted merging the above row-first-decryption-result andcolumn-first-decryption-result. All these results can be used to carry on inverse sparsetransformation and get the final decrypted image, separately called separate-combinedecryption and separate decryption. The former combines the feature of two kinds of separate decryption, and gets a better decryption effect. Through the analysis andsimulation to evaluate the calculation complexity and decryption performance of CSimage encryption system, it shows that the system can ensure a high security androbustness, and can achieve a high real-time performance and comprehensibledecryption image quality.In general, the proposed CS encryption system has realized the integration functionof “encryption”,“compression”, and “anti-packet loss”, serving as a practicable andnovelty solution for effective, safe and reliable image communication.