Font Size: a A A

Research On Face Recognition Based On Particle Swarm Optimization

Posted on:2022-03-24Degree:MasterType:Thesis
Country:ChinaCandidate:Y H WuFull Text:PDF
GTID:2518306332495834Subject:Electronics and Communications Engineering
Abstract/Summary:PDF Full Text Request
With the progress of society,face recognition technology has been fully developed,and has been widely used in many fields,bringing huge social benefits.Face recognition is an attractive and promising subject.It is of great significance to study face recognition technology.In the article,the face recognition science is primarily learned,and the particle swarm algorithm is also studyed.With the exception of the benefits of ordinary algorithm,quick convergence and few parameters,the particle swarm optimization is also possible to drop into terre optimum or premature convergence.To solve these problems,an improved particle swarm optimization(PSO)algorithm is proposed and introduced into face recognition to improve the recognition accuracy and recognition rate.The major research work is as follows:(1)This article summarizes the related algorithms for face feature extraction,introduces several classic face databases,and discusses the main research contents of face recognition and the related knowledge of face recognition technology and research methods.(2)On account of the inadequate convergence of the fundamental particle swarm algorithm,also the trend to drop into the terre optimum during the optimizing procedure,the article prescribes a PSO algorithm founded on the active change of the studying factor or inertia weight,which enhances the parameters of the studying factor or inertia weight.So as to finish algorithm optimization.With the continuous iteration of the algorithm,its learning factor and inertia weight are dynamically optimized as the number of iterations increase,and then balancing the global seek skill and the local optimizing skill.Experimental studies demonstrate that the enhanced algorithm is superior to the fundamental PSO algorithm on the basis of convergence rapidity or convergence exactness.(3)Since the traditional LBP algorithm is not good enough in feature extraction and the recognition rate is not high enough for face recognition,an improved LBP algorithm is proposed for feature extraction from Ar face database and ORL face database,the improved particle swarm optimization(PSO)is applied to SVM parameter optimization and face image recognition experiments are carried out simultaneously.In term of experiments,the effecting of the enhanced LBP algorithm has been enhanced.The SVM parameter optimization founded on the fundamental particle swarm algorithm is contrasted,the optimized SVM has the classification exactness,while it uses face recognition.Better,it is more efficient when used for face recognition.
Keywords/Search Tags:Feature extraction, Particle swarm optimization algorithm, Optimization, Face recognition
PDF Full Text Request
Related items