Gender recognition improvement: A new approach for extracting and selecting features

F. Maleki,M. J. Moghaddam,M. Moattar
DOI: https://doi.org/10.1109/ICTCK.2014.7033527
2014-11-01
Abstract:Human face based gender recognition is a challenging issue in image processing and machine vision domain. In this paper we proposed an approach for gender recognition using combination of statistical features and Local Binary Pattern (LBP). The optimal block size and statistical features set are determined by sequential forward floating selection (SFFS) algorithm for gender recognition improvement. The assessment and comparison with other methods have been carried out using Iranian facial image dataset. The proposed approach can carry out the classification more accurately. The rates of true classification using Support Vector Machine (SVM) and Multi Layer Perceptron (MLP) classifiers are 99.41% and 99.31 respectively.
What problem does this paper attempt to address?