Face detection using boosting and histogram normalization

Abstract

The aim of this paper is to apply the Viola and Jones face detection algorithm with the adjustment of using histogram normalization instead of variance normalization. Various sets of images were used to train and test the current approach. These sets were taken from the MIT Center for Biological and Computation Learning -CBCL- face database in addition to self-generated samples from the internet. The algorithm is applied to detect faces in gray scale images as opposed to colored images used in other algorithms. A comparison between the current approach and the original algorithm is presented.