Face detection using single cascade of customized features discriminator

Face detection is an important tool in human-computer interaction applications such as drivers assistant system that monitor drivers attention, which needs a head pose estimator that require multi-pose face detector. There has been a considerable amount of literature to address sub-problems of this problem, but the problem of multi-pose detection is still under study.;This thesis suggests a multi-pose face detection algorithm for uncontrolled environments that implements a single cascade of classifiers. Each classifier addresses a certain area of the problem. The design was aimed to maintain speed and an acceptable detection rate. The cascade implements fast and simple classifiers at first stages of the cascade.;Features were formed using facial features extracted by a knowledge-based filter and were variation reduced. Results on FDDB benchmark showed 5.22% detection rate with 2000 false positives, and on MIT+CMU testset showed 43.56% detection rate with 504 false positives.