A Novel Incremental Face Recognition Method Based on Nonparametric Discriminant Model

Face recognition has received considerable interest owing to its relevance in several domains. Yet, it has some difficulties in many real world applications, where data are collected continuously and must be updated over time. In the present, we advance an adept face recognition technique that rests on Incremental Nonparametric Discriminant Analysis (INDA). We use the Gabor wavelet to extract facial features on which multi-lob ordinal filter are applied to derive Ordinal measures that are encoded in local zones, as visual parameters. Then, the statistical dispersion of these parameters is integrated to get a feature vector whose dimension is decreased by making use of PCA and variance. Last but not least, every single feature vector is treated as a feature input for the INDA. The proffered face recognition technique was assessed on the well-known ORL and Yale face databases. Experimental results have shown clearly its superiority and skillfulness in terms of recognition performance.