Research On Video Data Security Technology Based On Chaos

With the rapid development of mobile communication technology and multimedia technology,video services and video applications have become an indispensable part of people's production and life.As an important carrier of multimedia information,video has been widely used in many fields of information such as military secret communication,video on demand,video conference and monitoring system.However,video is vulnerable to various attacks during storage and transmission,resulting in a series of security issues.Therefore,encryption protection of video data is a necessary technical means to promote the development of video applications.Video has the characteristics of large amount of data,strong real-time performance,and complex coding structure.Traditional cryptosystems are not suitable for encrypting video data.Chaotic ciphers have the advantages of strong key sensitivity,fast encryption speed,easy deployment and implementation.Therefore,using chaotic ciphers to encrypt video data provides a new idea for the protection of video information.The amount of video data is huge and must be compressed and encoded to be stored and transmitted.Therefore,for the encryption of video data,whether it is combined with the compression encoding process or the compressed code stream,the compression encoding format,compression efficiency and time overhead of the video must be considered.The low-dimensional chaotic system is fast and efficient and it is suitable as a cryptographic system for video encryption.Based on chaos theory and video coding theory,this paper constructs chaotic ciphers by using extended unimodal chaotic system,and video encryption algorithms that meet the requirements of security and real-time requirements are designed for the uncompressed original video and the compression-encoded video stream.The main research work is as follows:Firstly,the cross-compound unimodal chaotic system and homogenization algorithm is proposed in this paper.Low-dimensional chaotic systems such as Logistic system have fast iteration speed and high execution efficiency,so the time complexity is low,which is especially suitable for constructing stream ciphers to encrypt video data with high real-time requirements.However,low-dimensional chaos generally has defects such as uneven track distribution and narrow chaotic parameter interval.Therefore,a new unimodal chaotic system is proposed with constant Lyapunov exponent,scalable phase space,wider range of chaotic parameters and no periodic window.A general phase space homogenization algorithm is proposed for the unimodal chaotic system,which solves the problem of phase space boundary aggregation.In order to further improve the nonlinearity and randomness of chaotic sequences,the unimodal chaotic system is combined with the extended sinusoidal mapping to form a cross-compound unimodal chaotic system.The Lyapunov exponent of chaotic sequences generated by the cross-compound unimodal chaotic system is doubled.The randomness test index has also increased significantly,laying the foundation for designing a secure pseudo-random key stream generation algorithm.In addition,chaotic sequences are all in the real number domain,but the key stream is in the binary integer domain.Therefore,the chaotic sequence value must be transformed into a sequence of integers through the discretization process.Hence,this paper proposes a pseudo-random key stream generation algorithm based on linear perturbation and nonlinear coupling.Classical discretization methods,such as thresholding or segmentation mapping,cannot avoid the problem of short-period and chaotic degradation caused by computer finite precision.In this paper,the linear feedback shift register is used as a linear perturbation component,and the cyclic cycle is broken by bit recombination.The chaotic sequence is nonlinearly coupled with the bit recombination sequence,so that the finally generated key stream sequence has no precise period,and the anti-attack capability is also obviously improved,which satisfies the security requirements of the stream cipher.Thirdly,for the original uncompressed video file,a dynamic video encryption algorithm based on the cross-coupled spatiotemporal chaos is proposed in this paper.Encryption of the original YUV video is usually a combination of encryption and compression coding,and the compression and encryption processes are synchronized to save time overhead and computational resources.However,in order to ensure video format compatibility,only selective encryption can be used to extract some sensitive video syntax elements as encryption objects.Excessive encryption of syntax elements will result in longer encoding time,increased bit rate,and affect compression efficiency.Too few encrypted syntax elements will affect the encryption effect and reduce the security of encrypted video.Therefore,in order to solve the problem that the real-time and security cannot be balanced,all the video syntax elements which do not affect the compression efficiency and the encoding format constitute an encryption candidate domain.The objects for encryption are decided by cross-coupled spatiotemporal chaotic pseudo-random sequence dynamically.The selected objects are encrypted by the mathematical operation with key stream.Since the encrypted data in each frame is randomly changed,security is further enhanced.In addition,the real-time and format compatibility is ensured.Fourthly,to protect the compression-encoded video information,a robust video encryption algorithm for compressed code stream is designed.The encryption of the compressed coded video stream must be performed on the basis of fully parsing the code stream structure.The encryption operation should avoid the parameter information related to the decoding,otherwise the decoding will fail and the video cannot be played.In this paper,the H.264 code stream is analyzed in detail,and finally the data payload part of Slice is determined as the encryption operation object,which does not affect the decoding and can protect the video content.In addition,the existing encryption algorithm for the video code stream usually adopts the stream cipher system,and both the sender and the receiver use the same key stream to implement the encryption and decryption operation.However,once packet loss occurs during transmission,this method will cause the ciphertext and decryption key to not match,resulting in decryption failure.In order to solve this problem,this paper proposes a synchronization vector mechanism.The head of the network abstraction layer unit and the slice header are extracted to calculate the synchronization vector for controlling the chaotic system,so that the decryption key stream can always match the received ciphertext.The robustness of video encryption and decryption is solved by this mechanism.The synchronization vector is related to the plaintext information,so the key stream controlled by the synchronization vector is also related to the plaintext.Therefore,the video encryption algorithm can resist the selective plaintext attack,and the security of the video encryption system is further improved on the premise of meeting the robustness requirements.