The Improvements And Applications Of Feature Extraction Method Based On Module Image

Face recognition continues to be a hot topic in pattern recognition field due to its wide range of applications such as distinguishing somebody from others and controlling authority.This paper use the methods such as PCA,LDA,Difference Space,MSLDA based on module image feature extraction to fuse.The main work reads as follows:(1) This thesis introduced the background and scale and methods of the research, the latest development of face recognition methods,and the difficulties that exist.(2) The paper investigates the Principle Component Analysis(PCA) approach deeply,and find the advantages and disadvantages of PCA approach.But 2DPCA can take less time on calculation.(3)The methods based on fisher discriminant analysis for face recognition are investigated.The traditional LDA method often suffers the small sample size problem, so many new methods are proposed to solve this shortcoming,and PCA+LDA is one of them.Then a new method,proposed-2DPCA+LDA is also investigated.The experiment shows that 2DPCA+LDA is better than traditional PCA+LDA and 2DPCA.(4)Modular image based on PCA+LDA is analyzed deeply.Firstly,we introduced modular PCA method.In this method,the original images are divided into modular images,which are called sub-images.It is proved simpler and better than that of other technologies such as traditional PCA through experiments.On this basis,this paper prompt the new methods such as MPCA+MSLDA based on module image,MPCA based on Difference Space,MPCA with weight.Through experiments,we verify that the three new method are better than the traditional MPCA on the recognition rate,and have high robustness in lighting conditions.(5)Developed a face recognition system based on VB platform.The system has two parts:First,it use the selected feature extraction algorithm to experiment on the face database to obtain the recognition;Second,the real-time system can capture the face from the camera and decide if the face sample is belong to the person in the database or not.It can also add the new samples captured from the camera to the database.