Electrooculography (EOG) is one of the occulography methods used for the estimation of eye orientation. These signals, generated by eye movements, can be used in an efficient way as input in different control systems. So, the signal processing of the EOG signal is a key point when performing complex tasks, for instance, in a Human-Machine Interface (HMI). In this sense machine learning algorithms allow patterns in data to be identified, and then, to predict future actions using those patterns that have been learned. This paper presents a signal processing module for EOG signals, applying Wavelets Transform (WT) as a denoising procedure and AdaBoost as a machine learning algorithm.