Facial Feature Crop Detection Based on Haar Feature and AdaBoost Training

Abstract

The advance of technology has created a demand on face detection, which is the first phase in face recognition, one of the important biometric technologies. In this research, the Facial Feature Crop Detection Based on Haar Feature and Adaboost Training framework is presented. The framework’s objective is to detect face crops and each facial feature’s crop like eye, mouth and nose crops; the framework depends on using Haar-like features and the AdaBoost learning algorithm. This algorithm has been chosen according to its computational effectiveness and easiness. The presented framework was tested on a dataset containing 500 colour images of faces collected from the FERET database. The training achieved 100% accuracy and 98.8% achieved for the testing set.