Design Of Identity Recognition System Based On Lip Characteristics

With the development of research on biometrics,such as voice recognition,face recognition,iris recognition,etc.They all have been widely used,and lip feature-based identity recognition is a branch of biometrics,which separates individual differences and pronunciation.The combination of individual differences and pronunciation is more difficult to imitate,easy to maintain,and easy to use.This article analyzes and summarizes the current lip feature extraction problems,and proposes a lip feature extraction scheme based on the current research.The work is mainly divided into the following four parts:Firstly,the paper designed an overall identity recognition system based on lip features,the entire system is divided into a video pre-processing module,the annotation module key feature,a lip feature extraction section module and the identification module.Secondly,the video pre-processing module is subdivided into image framing,image cropping,image extraction and other parts,and the paper designed algorithms of each part.Thirdly,in the feature extraction part,in order to reduce the complexity,the key frames are filtered based on the time series changes of the frame image after the image is framed,and the lips are based on the six key feature points.Features are divided into geometric features and motion features.The curvature change of the lip curve of the speaker at different moments is obtained by fitting six key feature points;the opening and closing angle of the speaking point is constructed by connecting the key points,and the angle change at different moments is obtained by mathematical methods.Combine them to form lip features.Fourthly,based on the extracted lip features,an identity recognition system is designed through a similarity calculation algorithm.And through comparative analysis,the effectiveness and rationality of the system are verified.