Design And Implementation Of Match-Making System Based On Iamge Search

In this age of "physical appearance and finally soul," looks play an important role in the dating process.Although dating network provides a good platform for single people dating and making friend,the dating web search system still exist deficiencies,resulting in users' search results may not be ideal.The principal causes of these problems is the current dating network system only allows user to search according to certain keyword and cannot according to contain a face image.In order to improve the above problem,this paper designs and implements a dating web system based on face image search technology.This thesis starts with researching face recognition related algorithms,and utilize these algorithms to design and implement search system of match-making website,then implement match-making website based some open source libraries and open frameworks.The main work of this thesis is:1)Analyzed the functional requirements and non-functional requirements of the dating web system and designs the overall architecture of the dating system.The system adopts the design pattern of MVC,which makes the system clearly hierarchy,and the decoupling of the system is realized.2)Studies the support vector machine(SVM)algorithm,and by using artificial swarm algorithm(ABC)and particle swarm optimization(PSO)algorithm to optimize the related parameters of the S VM.3)Built a convolution model,and regard SVM as its classifier.And treat convolutional neural network model as the core component of face image feature generated model.4)Implements the face image feature model which bases on convolutional neural network module.In this model treat an image which contain human face as input parameters,and then using convolutional neural network model generated face image feature vector.All above work is to get better understanding of image information.5)Design and implementation the face image search module of the dating web system to achieve the goal of searching for the desired heterosexual user through images that contain faces.