Appearance based gender classification with PCA and (2D)2 PC A on approximation face image

Gender classification can play a significant role in security and surveillance system. It aids in identification of a person by recognizing its gender (male/female) from the face image only. Extracting discriminate features for male and female is a fundamental and challenging problem in the field of computer vision. In this manuscript, a combination of Approximation Face Image (AFI) with Principal Component Analysis (PCA) and (2D)²PCA (Two-Directional Two Dimension Principal Component Analysis) are applied for feature extraction, and SVM is used for classification. The experiments are conducted on a number of well-known face image databases taken in controlled (FERET, FEI, AR, and Indian Face) as well as uncontrolled environment (LFW). The experimental analysis shows that the proposed approaches give an acceptable classification rate with reduced feature size. The results also show that the proposed (AFI+(2D)²PCA) approach reduces the computational time as compared to AFI+PCA.