|
Face recognition is an important field in the pattern recognition and computer vision reseach, it has important application value in the public security, intelligent control, identity authentication, digital multimedia and e-commerce. In recent years, face recognition has been widely used in many aspects, the practical requirement of face recognition is more and more strict.Face detection is an important link in automatic identification system, face detection problem derived from the face recognition. Face detection is the prerequisite to rapidly and accurately identify facial condition, its purpose is to detect the face from the background image.In color images, color of skin is an important information of face. It dosen’t rely on the details of the facial features, and it has stable function and good effect, etc. In the chromaticity space, complexion has certain cluster. Using complexional characteristics, it is rapid to abandon a lot of background factors, significant to reduce the search space and can improve facial speed, so it can get the target region.The main research contents of this paper are:1.This paper briefly introduces the concept of face detection, application and domestic and international research situation, then compare and analysis several kinds of common color spaces, select the YCbCr color space which has cluster for complexion and establish complexion model, then calculate similarity and do binary. In order to eliminate noise, and then,based on binary, deal with the noise problem using the filter of digital morphology,to get an approximate face candidate region.2. In this paper,determine the stage of making sure selected face region, use method of projection and face geometry approach to effectively detect the human face.This method workshop well,fast and accurate detection of human face.3.Finally, this paper has studied an important organ of face--eye,it is extremely important in facial features.However,for a fast,efficient and stable algorithm for precise positioning of eyes is very important.For the positiong of the eyes to make the following work:the use of principal component analysis to extract the eye image features will be recognized,a man eye extraction efficiency of the method,and the external environment has a strong anti-interference.The method of positiong,high efficiency,the background size and other details have good adaptability,real-time monitoring system in the human face has a good value.
|