Research On Particle Swarm Optimization Algorithm For LDA Face Recognition Algorithm

In modern society,computer technologies is developing rapidly,in particular,the internet is getting more and more important,and all kinds of information may have the risk of exposure to the internet,therefore,it caused more attention by people for about how to ensure the information security during the process of using a computer and surfing the internet.As an emerging and evolving technology,the face recognition has a series of advantages as intuitive visualized recognition mode,reliable identification,concealed recognition process and so on,so that it became a popular researching object in the field of modern computer information security.This paper propose a new face recognition system based on PSO and LDA algorithm,the main research contents are as follows:The principle of the face recognition and the detailed introduction of the process of image pretreatment has been presented at the beginning of this paper.The process mainly includes the basic concepts of graying of color image,binary and so on,and the standardized processing of image,enhance the image effect,reduce the external affection to face image recognition quality and so on.Afterwards,analyzing the defect of using LDA algorithm during the process of face recognition,its shortcoming reflected mainly in three parts:(1)Small ample size problem due to the sample shortage;(2)Neglect the feature of sample itself while managing the sample data but adopting the same criterion functions;(3)While managing the complex variance matrix,it occurs no results most of the times.To avoid the defects above,this research attempted to optimize the original algorithm as to making the face recognition system more accurate by merging the particle swarm optimization(PSO)into LDA algorithm.Using the PSO algorithm to proceed the identification of variable randomly and using the LDA algorithm to evaluate the recognized variable,distinguish the most appropriate variable as the specific information of face recognition after the combination of this two algorithm.In addition,the other algorithm that can solve the three defects above are introduced.In the end,the PSO-LDA algorithm used in this paper has been compared with the original algorithms in ORL face database through experiments,and the results proved that the method adopted in this paper is advanced,and it could recognize the face more accurately.On this foundation,the face recognition system has been established in MATLAB,using self-built face database for testing,the results shows that the speed of recognition is satisfactory and the real-time requirements can be reached.