Hybrid Algorithm Of Face Detection Research

Face detection is to locate the faces and determine their size, position and amount in an image. As the first step of person identification, human-computer interface, intelligent scene supervision system, facedetection has been paid much attention and became a very active research branch pattern recognition and computer vision application areas with its signigcant value in recent years.This thesis is focused on arbitary backgtound image which maybe contain human face and combined with the skin color segmentation, candidate region by facial characteristic, Gabor wavelet feature extraction and back propagation neural networks.This article first introduced the background and present situation of face detection, and expounded the human face detection technological development vital significance. It has carried on the summary to the present commonly used some detection algorithm, then elaborated emphatically based on the skin color segmentation, candidate region by facial characteristic, the Gabor wavelet feature extraction and the neural network classification recognition of the human face detection algorithm.The skin color is the human face important characteristic. In after the skin color sampling statistics and the region cluster analysis. It establishes one kind under the YCbCr space skin color segmentation method, and completes the skin color region under the best threshold value selection algorithm the segmentation, obtained the kind of skin color region. Again to the skin color region which obtaims carries on human face characteristic screening by the image geometry attribute and the analysis situs attribute, further rejected the inhuman face skin color region, reduced the candidate face quantity, simplification following detection process processing. This paper uses the Gabor wavelet to carry on the feature* extraction for the candidate face, and carried on the classified recognition through the BP neural network, enhanced the accuracy which the human face detection.Finally, the algorithms in this paper can detect multi faces with multiple orientations a complex background. The experimental result indicated that the algorithms are effective. It has the high detection performance and the low miscarriage rate.