The Experimental Research Of The Real-time Face Recognition System Based On The Embedded Platform

Face recognition is a hot research topic of the recognition pattern,computer vision and image processing.The application of face recognition is also popular in the market.The research on the traditional face recognition is usually implemented with the help of the computer technology.With the rapid development of the embedded technology,the embedded platform is increasingly well-accepted by more and more face recognition researchers due to its advantages of systematism and conciseness,reliability and specification etc.,setting off a wave of research and application boom about the embedded system based on face recognition.Based on the embedded hardware platform,this paper designs a new face recognition system with a high recognition rate and a fast recognition speed.With the aim,the work is divided into the following parts.Firstly,an embedded hardware platform is set up.And the experiment selects the popular Raspberry Pi development board based on the ARM for use.Compared with the one integrated with ARM and DSP,the development board has a superior effectiveness;Additionally,compared with the development board based on ARM9,it has a more powerful computing performance.And the experiment chooses the CSI interface camera officially released by Raspberry Pi as the image acquisition equipment.Besides,an Linux-based Raspbian operating system is installed on the development board,and an OpenCV computer vision library is transplanted so as to achieve a better specification.The designed experiment is based on the embedded platform to achieve the face recognition.In this paper,the author firstly makes a brief introduction to the image preprocessing steps including the image format conversion,the coordinate transformation,the correction of geometric distortion as well as the histogram equalization.Then,on the basis of image preprocessing parts,the principals and applications of the face motion detection module,the face detection module,the face feature extraction module and the face recognition module are introduced and explained in detail.The face motion detection adopts the differential image method and the face detection module is fulfilled on the basis of the Haar + AdaBoost algorithm.After comparisons are made among the PCA method,ICA method and LDA method,the face feature extraction selects the PCA method to realize the extraction of facial features.What's more,the face recognition is achieved with the use of the nearest neighbor classifier.Finally,the entire system is tested.The result of the test shows that the face recognition based on the embedded hardware platform can achieve the design expectation,and has a relatively higher recognition rate as well as a faster recognition speed.And the experimental result has certain application values and reference values.