Encryption of text file using a user controlled automatically-generated key

Traditional symmetrical cryptographic algorithms generally provide an adequate degree of immunity to attacks aimed at revealing secret keys. A number of approaches exist for the automated generation of secret keys, but, for high security applications, some end users remain wary of approaches that are controlled by third parties. Consequently, there remains interest in certain high-security applications in being able to retain control over the method used for the generation of keys. In this paper, keys for both image encryption and decryption are obtained using the evolutionary computing tool Eureqa, in its modelling of pseudorandom input data. The secret keys generated by this approach and when applied to the encryption and decryption of gray-scale images are validated in a range of statistical tests, namely histogram, chisquare, correlation of adjacent pixel pairs, correlation between original and encrypted images, entropy and key sensitivity. Experimental results obtained from methods show that the proposed image encryption and decryption algorithms are secure and reliable, with the potential to be adapted to high-security image communication applications.