Research And Implementation Of Identity Recognition System Integrating Coercion Detectio

In recent years,biometric identification technology has been widely concerned,and people pay more and more attention to personal information security.Single-mode identification is vulnerable to external environment interference,and cannot meet the high-precision and high-security places,while multi-mode biometric identification effectively integrates two or more biometric features,which can make up for the shortcomings of single biometric identification instability and high error rate.Therefore,in this paper,the finger vein features and facial features are fused to build a high-security multi-modal identity recognition system,and considering the legal person may be stressed for identity authentication,the design of auxiliary stress detection of dual-modal identity system,focusing on image acquisition,image quality evaluation,stress detection based on expression recognition and dual-modal identity recognition four parts.The main work is as follows:(1)Aiming at the problems of low contrast,environmental noise and motion blur in the process of image acquisition,and aiming at the characteristics of near-infrared light acquisition of finger vein images,a comprehensive vein quality evaluation method was designed,which included contrast,information entropy,brightness uniformity,sharpness and noise intensity.An adaptive dimming circuit was designed for the thin front end and thick back end of fingers,and the near-infrared light intensity was adjusted flexibly by using the results of the above venous image quality evaluation index,to effectively guide the process of image acquisition,so as to achieve fast and high-quality image acquisition.(2)In view of the possible stress recognition behavior in user identity recognition,a facial expression recognition method integrating pyramid gradient histogram and Gabor was proposed to assist the stress detection.In this paper,a three-layer pyramid structure is designed to detect the edge and shape features of the facial image efficiently and to count the local facial information effectively.Combined with multi-scale Gabor filter to extract the frequency domain features of the image,filter out irrelevant face information,and then combine the above time domain features and frequency domain features to play a complementary role,so as to improve the accuracy of expression recognition,effectively identify the potential stress of the expression category,to achieve the effect of assisting stress detection.(3)Aiming at the problem that single-mode biometric feature recognition is easily affected by the environment and the recognition performance is unstable,a two-mode recognition method based on texture features is proposed.In view of the phenomenon that finger vein imaging is easily affected by light source and fuzzy image,a feature extraction method of finger vein was designed by combining local binary mode and local phase quantization,so as to achieve the purpose of being insensitive to light mutation and fuzzy invariance.Aiming at the directional characteristics of facial organ edge changes,a face feature extraction method was designed to integrate local directional finger and local phase quantization,so as to effectively extract the direction information and phase information of images.On this basis,the above two modal characteristics are fused at the feature level to construct a dual-modal recognition system,which can effectively reduce the error rate of the system and improve its anti-interference ability of the system.(4)Aiming at the problem of insufficient use of information in shallow feature recognition method,a dual-modal identity recognition method combining texture features and CNN is proposed.With texture feature map as model input,a VGGNet based identity identification network is constructed,which makes full use of the shallow texture information and deep feature information of images,and effectively improves the performance of dual-mode identity recognition system.