A hybrid improved kernel LDA and PNN algorithm for efficient face recognition

This paper proposes a hybrid approach to face recognition based on a combination of probabilistic neural networks (PNNs) and improved kernel linear discriminant analysis (IKLDA). The dimensions of a sample's features are first of all reduced, whilst retaining its relevant information, A PNN method is then adopted to solve face recognition problems. The proposed IKLDA+PNN method not only improves the overall computing efficiency, but also its precision. Face recognition experiments conducted on the ORL, YALE and AR datasets, which contain a wide variety of facial expressions, facial details, and degrees of scale, were used to validate the feasibility of the IKLDA+PNN method. The results showed that it can obtain an average recognition accuracy of 97.22%, 83.8% and 99.12%, across the three datasets, respectively. (C) 2019 Elsevier B.V. All rights reserved.