Waveletfaces and Linear Regression Classification for Face Recognition

The face recognition task has become one of the major research topics because of its applications such as biometric security and authentication. State-of-the-art methods for the task intend to maximize the classification accuracy of different persons by extracting discriminant features, achieving a dimensionality reduction. In this paper, we propose a combination of the Wavelet decomposition technique with the Linear Regression Classification Algorithm (LRC). We evaluate the proposed method in five different data sets and using seven different Wavelet functions. The experimental results show that this approach achieved an improvement up to 18% in mean accuracy rate if compared with the LRC method alone.