Precise eye and mouth localization
In literature it has been shown the strong correlation between the precision of the face localization step and the face recognition performance: even small localization errors make the recognition fail. Given the criticality of the localization step, it arises the necessity of having precise facial feature detectors, and objective measures for their evaluation and comparison. In this paper we present significant improvements of a previous method for precise eye center localization by integrating a module for mouth localization. The technique is based on Support Vector Machines trained on optimally chosen Haar wavelet coefficients. The method is tested on several public databases and the results are reported and compared according to a standard error measure. The tests show that the algorithm achieves a high localization precision.