Pixel clustering for face recognition

This reported work proposes a theoretical framework for unsupervised feature extraction called pixel clustering. It is a simple, yet effective framework, which allows to define feature extraction techniques by setting just two parts: a pixel clustering algorithm and a linear/nonlinear combination of the pixel into a cluster, in order to create a single feature per cluster. The pixel clustering framework also makes it possible to create simpler, computationally cheaper and more general new implementations of well known feature extraction methods such as Waveletfaces and Eigenfaces. Two feature extraction methods are implemented over the pixel clustering framework, which are tested over three face datasets. Test results are compared against the traditional Eigenfaces and others state-of-art feature extraction methods for face recognition. It is shown that the proposed method overcome state-of-the-art methods.