chapter  7
12 Pages

Real-time face detection

A real-time face detection framework works in situations where the processing of face images is carried out extremely rapidly with high true detection rates. The most prominent system which achieves such an objective is known as the Viola-Jones face detector [112],[113] and the material of this chapter is mainly based on his works. In the majority of cases the technique is implemented in wide ranges of small,low, power devices, including handheld and embedded processors. This extremely fast face detector has broad practical applications where rapid frame-rates are not necessary. These include user interfaces, image databases and teleconferencing. Though the training is slow in this system, the detection rate is very fast. This face detector has three main ideas: (1) new image representation called integral image allows for very fast feature evaluation, (2) a simple and efficient classifier that is built by selecting a small number of important features from a huge library of potential features using an algorithm called AdaBoost and (3) a method for combining successively more complex classifiers in a cascade structure which dramatically increases the speed of the detector by focusing attention on promising regions of the image.