Local Binary Patterns for Gender Classification

Abstract

Several approaches for gender of handwriting are proposed an appearance feature-based approach. In this paper we present a comparative study to evaluate effectiveness of different Local Binary Patterns methodologies in characterizing gender from handwriting. We investigate different local binary patterns (LBP) parameters with/without preprocessing step based on low-pass filtering as features to represent handwriting images. Features extracted from male and female writings are used to train an SVM. The system is evaluated on the standard QUWI database depending competitions of ICFHR 2016 of handwriting images and realizes promising classification rates.