ROBUST GENDER CLASSIFICATION USING LMnP - LOCAL MINIMA PATTERN
Mohammad Shahidul Islam · 2013
Gender classification is an important matter for Human Computer Interaction devices. A new methodology for gender classification is examined in this study where the facial feature is extracted from local region of a face using gray color intensity. The facial area is divided into eighty-one equal sized square sub-regions and Local Minima Pattern (LMnP) method is applied to each pixel. LMnP histograms extracted from those regions are concatenated into a single vector to represent that particular face. The classification accuracy obtained using Local Minima Pattern (LMnP) along with support vector machine as a classifier has outperformed all traditional feature descriptors for gender classification. It is evaluated on the images collected from popular FERET database