Abstract—The development of face recognition system attracted significant research attention due to increasing demand on its applications. However, the process of converge to a conclusion of a known-face based from a 2D incoming face images is very difficult, especially when large illumination variations are present in the input space. In this paper, we implemented the illumination compensation praprocessing system in conjuction with the optimized-Probabilistic Neural Networks as a classifier. PNN has shown marvelous higher recognition capability, however, determining the best tuning parameter is very difficult. Neural topology is firstly determined by looking for the most representative neuron using Orthogonal Algorithm, followed by evolutionary determine the best smoothing parameter through Genetic Algorithm. Experiments are conducted using face images under various illumination conditions, and results are presented which illustrate the potential of this approach.

Index Terms—Probability neural networks, genetic algorithm, orthogonal algorithm, illumination compensation, face recognition.

B. Kusumoputro is with the Electrical Engineering Department, Universitas Indonesia, Depok Campus, Depok Indonesia (e-mail: kusumo@ee.ui.ac.id).
Lina is with the Department of Computer Science, Faculty of Information Technology, Tarumanagara University, Jakarta, Indonesia (e-mail: lina@untar.ac.id).

Cite:Benyamin Kusumoputro and Lina, "Invariant Illumination Face Recognition Using Optimized PNN Topology through Neuron Orthogonalization and Genetic Algorithm," International Journal of Computer and Communication Engineering vol. 3, no. 1, pp. 31-36, 2014.