Research Of Face Database Optimization Method And Application Based On Multi-view Image Data Mining

In recent years,electronic commerce as a new form of business,has gained in-depth development.In order to ensure the safety and effectiveness of electronic commerce transactions,the identity of the user authentication has become a very important research problem in the field of electronic commerce.At present,the face recognition system is used to verify the e-commerce user's identity information,which is widely accepted because of its unique information,moderate distance and high security.The face database,which as the key component of the face recognition system,the retrieval performance and standardization degree of face database affects the face image retrieval speed and face recognition accuracy directly,so face database optimization has important research significance for enhancing the security and effectiveness of electronic commerce identity verification.In order to solve the problem of slow retrieval speed and low standardization degree caused by complex factors in the face image database,a new theory and method of face database optimization and application research based on multi-view image data mining is proposed.The main research contents of this paper are as follows:Firstly,a method of face database optimization based on compound gradient vector model is proposed,which is by imitated in the direction of biometric system to face feature spatial image perception,to realize the "bionic"optimization of face database in airspace perspective;Secondly,a method of face database optimization based on frequency cluster model is proposed,which is based on the characteristics of the information entropy and the gray energy can maintain a stable overall feature distribution under complex background interference,to realize the "active" optimization of face database in frequency domain perspective.Thirdly,a method of face database optimization based on 3D perspective is proposed to realize the rapid optimization of 3D face database.Finally,according to the application characteristics of the electronic commerce authentication system,an electronic commerce authentication scheme and its system model based on multi-view face database optimization is proposed to verify the effectiveness of face database optimization.The innovation of this paper mainly includes the following three aspects:(1)The method of face database optimization in spatial domain captures the characteristics of the biological vision system which has the natural perception and the ability to capture details.The spatial image features are described by a composite gradient vector model with independence and restriction.It can overcome the interference of complex factors on the face database optimization effectively and improve the retrieval speed and standardization degree of the face database;(2)The method of the face database optimization in frequency domain captures the image frequency information which can effectively represent the distribution probability of the overall appearance information of the target image.It is very effective to optimize the face database under the influence of complex factors.At the same time,the decision tree model is constructed by using the prediction information of frequency cluster model.It can effectively overcome the interference of the image feature deformation on face database optimization,and improve the retrieval speed and standardization degree of the face database;(3)3D face database optimization method captures the characteristics of 3D face image information dispersion and large amount of information.Three-dimensional face image information is characterized by three-dimensional singularity and three-dimensional feature points with stable data structure,low information volume and compact structure,which is making the model more accurate and reliable to ensure the optimization performance of 3D face database.