Automatic Face Recognition Based 0n Metaheurestic Key-Point Selection

Abstract

This article brings into view an automatic face recognition approach based on a metaheurestic Modified Shuffled Frog Leaping algorithm and machine learning techniques. The proposed method takes advantage from classic Gabor filters in order to extract and select the most salient areas in the face shape. Although most of researchers are attempting to seek a greater parameterization for the classic Gabor wavelets, this work focuses on enhancing the exploitation of these potential filters based on multiple machine learning tools. Thus, the adopted system investigates the impact of blending deep filtering with the metaheurestic behavior to achieve key-point selection. The convergence of the evolutionary memetic steps to an optimal criterion permits to upgrade Support Vector Machine classification results and reach the highest recognition scores. The effectiveness of the proposed face recognition system is validated by carrying out four common databases based on Radial Basis Function. Experimental results demonstrate that our approach outperforms other existing methods in terms of precision and positive recognition.