A ROBUST HUMAN FACE DETECTING METHOD IN COMPLICATED BACKGROUND IMAGE

摘要

A robust human face detecting method in complicated background image. It adopts microstructure features with high efficiency and high redundancy to express the gradation distributed character of the eye, nose and so on of human face mode and adopts the cost sensitive AdaBoost algorithm to choose the microstructure features of human face and non-human face to form a strong classifier, so that each layer classifier can reduce the false acceptance rate of non-human face samples in the case of ensuring extremely low reject-recognition rate, therefore realizing human face detection of higher performance in complicated background image with easier structures. Additionally, post-processing algorithms are used to further reduce the misdetection rate.