In recent years, researchers have been eager to improve image security via various trustworthy encryption models. Normally, image encryption is built on a single PRNG generator which can be considered vulnerable allowing security weaknesses. This research proposed a different image encryption algorithm based on random selection of combining types of efficient PRNGs to further increase the security. Our proposed system focuses on effective keystream generation by using a binary gravitational search algorithm to select random PRNGs starting encryption key followed by chaotic gravitational search algorithm to produce the initial value for the selected generator. The work uses the developed keystream to encrypt images utilizing smart permutation and substitution processes. The experimental results showed acceptability of the proposed method over others. The work evaluation of several statistical measures provided interesting remarks very much pioneering opening interesting directions for future research to come.