Pattern recognition is a mature field; however, in recent years it has developed a special interest. Many methods have been developed exclusively focusing on this subject; all of them, however, are centered on certain types of patterns. In this paper, a new method, which aims to be as flexible as possible, is proposed. This technique is based on a validation standard method used in electromagnetic compatibility named feature selective validation. The final objective of developing the algorithm is to make it adaptable to diverse types of patterns. An application on the recognition of transient patterns is exposed as an example. Nevertheless, a wide range of other types of signals is likely to be treated with the same logic.