Research On Facial Matching Method In Attendance System

Biometric face recognition technology as a technology,it has a direct,friendly, convenient and easy for users to accept,it has become the best basis for authentication. Attendance is the basis of modern enterprise management,and the important indicator of the enterprise management level.The design of employee's time and attendance system based on the authentication of face recognition is enable to solve the traditional way in which IC cards frequent attendance of Doug card issues to ensure the authenticity of the data,truely embody open,fair and impartial.First of all,the subject of this article summarized the background and significance of the study and analyze the domestic and international research on face recognition and progress.on this basis,purposing the sub-leval rcogonition bsed on the weighted template matching and SVM,and combining image acquisition and access control subsystem to realize the design for the attendance/access control subsystem by layered human facial matching.As follows:(1) Designing and Implementing a color model based on two-dimensional projection of the face segmentation algorithm quickly and at the same time design a face image normalized,normalized scale and gray balance,and lay the foundation on the human face feature extraction and recognition.(2) Proposing Extraction Method for Face based on Gabor Wavelet Feature.The method adopted for the separate and normalized face image after the first of its grid to deal with,and so each grid Gabor wavelet transform,transformed into a feature vector,as a face image in the network Georgia Department of characteristics,which characteristics of the composition of face mask elastic.Extracting human facial feature vector based on Gabor wavelet transform,which can improve systems' tolerance of illumination and psture chnges(3) Proposing sub-level face recognition method based on template matching and SVM-weighted.The method set threshold and the two match,c₁ and c₂.Weighted template matching,if the Euclidean distance between the test face sample and Face template library card number,s meets s＜c₂,matching success;if c₁≤s≤c₂,the match results are not identified,move on to the next level to make the SVM classifier to identify,then determining the final outcomethan;if s＞c₂,then refused.The method is fast to identify and correct identification;at the same time it has high ability of outside library staffs rejection.(4) Considering the entire face attendance system hardware structure and focusing on an analysis of the function on the attendance based on the face subsystems,while using of object-oriented design methodology,designing and implementing the attendance in the face than on the prototype subsystem.