We present an efficient image authentication scheme using Genetic algorithm (GA). Using the crossover and mutation process of GA the original image is randomized into a binary string. A pairing function is then used as a checksum function that converts the binary string to a fixed length digest of the original image. A random permutation is used as the secret parameter between the authenticator generator and verifier. The scheme provides a non-reversible compression and collision resistance property, and is secure against chosen plaintext attacks. The experimental results show that the proposed scheme is efficient in comparisons to standard cryptographic authentication algorithms.