Further improvement of the HILL method applied in image encryption

In this article, a new encryption system that takes as an input an arbitrary size color image (n, m), and generates an encrypted image of the same size at the output, will be proposed. On each three-pixel block, an improvement of the classical HILL method provided by a size matrix (3, 3) invertible in the $Z/256Z$ ring, accompanied by a chaotic translation vector. Then a confusion with the chaotic map is established with the output vector. Finally, a chain with the clear block is performed to increase the avalanche effect. Simulations performed on a large number of images of different sizes and formats, show that our approach can withstand all known attacks.