There are several optical sensors available that can be integrated into a small satellite for the remote recording of different places on Earth. Colombia in its FACSAT-2 satellite mission has selected a multispectral optical sensor of medium resolution for this task. However, the associated data processing model does not allow the generation of useful products for end users, given the effects that phenomena such as the presence of clouds could cause in 8-band multispectral images captured by FACSAT-2. Therefore, this work proposes the possibility of establishing a real-time system for on-board processing of the data recorded by the satellite optical sensor, taking advantage of the processing resources offered by the embedded computer system in charge of integrating the payloads to the satellite bus and initiating the development of software applications based on artificial intelligence techniques, so that together they can meet the need for autonomously estimating the percentage of cloud coverage in multispectral images prior to the storage and download process, while complying with the restrictions imposed on this type of systems with respect to the time variable.