A classifier training method for face detection based on AdaBoost

In this work, we proposed two novel ideas for improved Adaboost-cascade face detection. Firstly through researching the characteristic of weak classifier, we proposed a method of computing threshold which obtained high detection rate for using fewer weak classifiers. Secondly selecting discriminative weak learners to optimize the detection performance and employing the number of Haar-like features in the Adaboost training. This approach maintains the simplicity of traditional formulation as well as being more discriminative. Mostly it is more efficient and a robust detector with few features. Simulation experiments in most static face detection and a little video face detection system are conducted that including human frontal faces and clutter, our method is superior to conventional AdaBoost in computer efficiency and increase the detection accuracy of the existing classifiers.