Abstract
Huge amount of sensitive data transmit over networks every day, cryptography is famous and widely used technique to prevent these data from unauthorized people; encryption/decryption strength depends on the algorithm and the key(s). Keys management including generation, goodness and distribution are very important and of the concern of the researchers. This paper focused on the keys generation from images data and highlighted the good features of the image that can be used for keys generation. A proposed algorithm for keys generation from images has been developed, three randomness tests: frequency/mono-bit, serial and poker tests have been implemented. The algorithm has been used to generate large number of keys of different lengths (256-bits and 512-bits) from variety of images with different features, the generated keys have been evaluated for randomness quality, results have been classified according to images features (entropy value and number of colors) and then analyzed and concluded. By conclusion, images data are not a good choice for keys generation, but some images that have a large number of colors and big entropy value can be used for this purpose.