Chaos-based image encryption schemes have been extensively employed over the past few years. Many issues such as the dynamical degradation of digital chaotic systems and information security have been explored, and plenty of successful solutions have also been proposed. However, the impact of finite precision in different hardware and software setups has received little attention. In this work, we have shown that the finite precision error may produce distinct cipher-images on different devices. In order to overcome this problem, we introduce an efficient cryptosystem, in which the chaotic logistic map and the Galois field theory are applied. Our approach passes in the ENT test suite and in several cyberattacks. It also presents an astonishing key space of up to 24096. Benchmark images have been effectively encrypted and decrypted using dissimilar digital devices.