Performance Comparison of Hybrid Methods on Facial Detection

Abstract

This paper surveys, reviews, and compares performance of facial detection using two hybrid methods: PCA+EBGM and PCA+LDA. The results of these hybrid methods provide data which gauges different performances. In PCA+LDA, linear combination for comparing and extracting classifier of data images into parts with input images and training images is used. In PCA+EBGM, the same linear method as was used with PCA+LDA—such as the incident light, the position of the face, and expression—cannot be used. Because PCA+EBGM using eyes alignment must be carefully considered, so that the method is highly accurate. Moreover, two points can be used for this method. However, this hybrid is faster, more precise, and will be more effective for greater amounts of data.