Patch-based Local Phase Quantization of Monogenic components for Face Recognition

International audienceIn this paper, we propose a novel feature extraction methodfor Face recognition called patch based Local Phase Quantizationof Monogenic components (PLPQMC). From the inputimage, the directional Monogenic bandpass components aregenerated. Then, each pixel of a bandpass image is replacedby the mean value of its rectangular neighborhood. Next,LPQ histogram sequences are computed upon those images.Finally, these histogram sequences are concatenated for constitutinga global representation of the face image. Using theproposed method for feature extraction, we construct a newface recognition system with Whitened Principal ComponentAnalysis (WPCA) for dimensionality reduction, k-nearestneighbor classifier and weighted angle distance for classification.Performance evaluations on two public face databasesFERET and SCface show that our method is efficient againstsome challenging issues, e.g. expressions, illumination, timelapse,low resolution, and it is competing with state-of-the-artmethods