Novel Methods For Encrypting Images Based On Blind Source Separation

Blind source separation image encryption is a reliable,high security of image encryption methods.Blind source separation is mixed signals on the basis of mutual independence between the source signal separation,finally got the source signal is estimated.Independent component analysis(ICA)is a recently developed a kind of blind source separation of signal processing technology,the application of the ICA algorithm for image decryption,blind source separation of this uncertainty may bring the decryption process is difficult,therefore,through the blind separation of image encryption methods also will be affected by certain constraints.This paper mainly studies are as follows:(1)summed up three kinds of typical algorithm of ICA,ComonICA,Jade,FastICA and Kernel_ICA algorithm,and introduced the basic knowledge of the information theory and separation principles,mainly including the definition of independence,moment and cumulant,kurtosis,information maximization method,minimum mutual information method,etc.(2)Jade algorithm is applied for image decryption,reference signal with noise as a secret key is proposed.Matlab simulation are used to realize the Jade algorithm.And the decryption time,similarity,SNR three performance indicators are analyzed.And then USES the traditional FastICA algorithm to decrypt the encrypted image,finally compared with the Jade algorithm,by comparing we found FastICA algorithm decryption of image stability is poorer.(3)Kernel ICA algorithm is used for three different core function,respectively image decryption,then noise as a secret key reference signal is used;Matlab simulation is used to realize Kernel_ICA algorithm.And the decryption time,similarity,SNR three performance indicators are analyzed.compared Jade algorithm with Kernel_ICA algorithm of image decryption,simulation verified the Kernel_ICA algorithm is effective and practical.