Research And Application Of OpenCV In Face Access System

Face detection and recognition is an important basic research topics in the field ofcomputer vision and pattern recognition.With the people’s awareness of securitystrengthening,basic research in the application of access control systems is becomingincreasingly important.This paper describes the development status of research and applications of face detectionand recognition technology at home and abroad, discuss the image preprocess proceduresbefore face detection and face recognition,and introduce a common face detection andidentification methods.Focus analysis an important mechanism of face detection: integration ofmultiple weak classifiers,Viola proposed real-time face detection algorithm based onAdaBoost,This method uses the characteristics of Harr-like to represent the image, theintroduction of the concept of the integral image, to improve the characteristics of calculationspeed,form a strong classifier using AdaBoost method to select a small number of features, usethe "Cascade" strategies to improve face detection speed, to achieve better detectionperformance.The article also elaborated Eigenfaces (PCA) face recognition method.Using K-Ltransform to extract the main component of the face constituting the eigenfaces space,in orderto filter out all the information account for the largest amount of information of the component(main component),eliminate minor constituents,Compare and evaluation the re-build the modelparameters by the correlation characteristic.It is the most widely used face recognitionbenchmark method currently. Focus on the advantages and disadvantages of face trackingalgorithm (Camshift),Propose improved algorithmuse of use AdaBoost algorithm for facepositioning,extract location information,and pass the result as the automatic initialization facewindow to automatic face tracking.OpenCV (Open Source Computer Vision) is an open source computer vision library whichsupport for Windows, Linux, Mac OS, Android, and other common operating system, codeportability.In this paper, some data types of OpenCV open source library such as CMat,IPlimage and OpenCV computer vision function interfaces are used,though the face informationobtained from the real camera,write a program to achieve human face detection and recognitionaccess control system. Finally, verify the face recognition rate and detection speed indicators,itis proved that the recognition credibility is82%and the average detection use-time is30ms.