Applications based on gestures or actions are increasingly common in everyday life. This is because it allows the user to activate the equipment without the need to touch it, besides being possible to activate it remotely. Thus, this work proposes a gesture recognition method using Principal Component Analysis (PCA) as a dimensionality reducer of variables, and the Euclidean distance as a classifier. The objective here is to interact with an environment through the actions of a user. To do this, an own database with 14 classes of actions were created. The classifier is validated through an online operation and a confusion matrix is used to analyze the results. The classifier achieved a hit rate higher than 80% for 6 of the 14 classes. This shows that the present work can be improved. However, it is worth noting that the main contribution was the formalization of a simplistic classifier, which makes use of a small database.