**7. Conclusions**

We presented an innovative method specially designed to enhance the decision performance for face recognition applications. This approach is based on a classification algorithm by means of nonparametric estimation of the regression function, which defines the probability of recognition. A two-step procedure was developed, considering first the construction of the correlation planes and then the decision-making based on a kernel smoothing regression algorithm. The results and their discussion show that this easy and fast to implement algorithm performs very well using the PHPID dataset. Our results are useful because we reach a very good level of recognition, namely less than 1% of MSE. This method can be extended in parametrically modelling correlation planes.

**Author Contributions:** Conceptualization, M.S., M.E. M.A., A.A. and C.B.; methodology, M.S., M.E. and A.A.; software, M.S. and M.A.; validation, A.A. and C.B.; investigation, M.S. and M.E.; resources, M.A.; writing—original draft preparation, M.S., M.E., M.A. and A.A.; writing—review and editing, M.S., M.E. M.A., A.A. and C.B.; supervision, A.A. and C.B.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.
