Research On Method Of Side Face Recognition

As the technology of face recognition,the side face recognition refers to extract characters of portrait of side face based on processing of side face image, and recognize the information about human face from side face database according to according face features matching algorithm. It’s one of most important topic which is in development and in the field of computer vision and pattern recognition, and is widely used in the field of security and protection、law and business. Because side face recognition technology lose more information,and side face characters are different from the positives,the detecting methods are different. This makes the side face recognition technology face many problems, and it needs more lofty people to study and develop to improve the effectiveness, stability and practicability.In the collection and analysis in recent years at home and abroad about positive and side face recognition technology on the basis of the literature,this paper propose to connect two technology’s characters which are technology based on geometrical features and technology based on curve fitting,and explore the two technology’s merit and demerit and optimize them. In this paper,first of all compare the procedure of pretreatment,and result in the image segmentation technology based on skin color model is fitting the technology based curve fitting and make flexible use of ostu technology to make it more exact; then compare the procedure of feature extraction,the procedure of the technology based geometrical features is more complex and sensible according to the technology based on curve fitting,and the procedure of the later is simple and stable.However the matching precision of the technology based on geometrical features is higher than the other because of the side face characters which can completely represent the face information.There are accurate matching and fuzzy matching. On the basis of the compare and analysis,the hierarchical matching algorithm has been put forward.The first matching is curve fitting and the result are used by the input of the second matching;the second matching that use technology based on geometrical features fit the images which are derived from the first matching’s result The hierarchical matching algorithm promote the recognition rate because therehave rejected many images which are difficult to identify.For there are many repeat parts between two technology,the according technology have been adopted to optimize the procedure of recognition.At the last,there are the experiments in this paper which contain the process and the result The experiments include three part:the experiment on technology based on curve fitting,the experiment on technology based on geometrical features and the experiment on hierarchical matching algorithm.The experiment results show that the hierarchical matching algorithm has greatly improved the recognition rate,practicabilityand effectiveness according to the technology based on curve fitting and the technology based on geometrical features.