Face Detection And Recognition Technology Research In Complex Background

Face detection and recognition technology is an important method for identity authentication based on biological characteristic, it has important application in the areas of safety access control,video surveillance, intelligent human-computer interfaces and content-based retrieve etc.This paper gives a review on the history of face detection and recogniton, then mainly discusses the face detection problems in static color image in complex background and varying illumination. It consists of four main parts: image-segmenting preprocessing based on skin color and frontal face detection based on eye gray, feature extraction based on PCA and face detection based on improved bp algorithm.First, study the clustering characteristics of face color under some color spaces , then choose YCbCr models for skin segmentation, establish a face Gauss model,make use of the Ostu algorithm improved to carry out skin check for face image,handle it in mathematics morphology. The experimental results show that the proposed approach is effective to detect human skin of any number with any poses in any background.Second, make use of border detection algorithm and work out face regulation, remove similarity skin and obvious inhuman face area. According to eye gray and eye area complicated degree characteristic, improve a kind of eye detect method, and normalizes, standardizes for face image. The experimental results show that the eyes can be detected by the proposed approach accurately.Third, make use of the PCA algorithm to extract face feature, get the characteristic-vector and the characteristic-value, carry out training and recognizing face according to the traditional and improved BP algorithm , The experimental results indicate that the improved BP algorithm can improve recognition rate and reduce train time farther.Finally, a fast detection and recognition system for images in the condition of complex background is proposed and implemented by Visual C++ and matlab, the model system accomplishes face detection and recognition from images in complex background robustly and accurately.