Design Of Iris Image Acquisition In Tracking And Pointing System

High precise characteristics iirs is suitable for identification authentication.Quickly capturing high-quality iirs image has been one of research hotspots in the fieldof iris recognition, only requiring user’s less cooperation in the big scene. Thelong-distance iirs image acquisition system mainly researches on three problems: irislocation and search, large depth of field and compensation light question. Currently,the limitation of iris location and searching speed mainly include two aspects, on theone hand，the PTZ is of slow rotation and low accuracy. On the other hand, thehardware cost and time overhead of the system increases by three-dimensionalpositioning of the face or the eye. Aiming at solving problems of iirs location andsearch speed, this dissertation a novel solution is proposed which used the eye trackingand relfector pointing systems. The system used reflector delfection driven by servomotor rotation to change the view of camera, instead of PTZ rotation. Face positionmethod in long-distance system is replaced by eye position, and reduced threedimensional eyes positioning to plane coordinate eyes positioning. The main contentsare as follows:1.According to specification and requirements of iris image, the relationshipamong each design parameters of iirs camera is analyzed in this paper. Configurationof the acquisition system of long-distance iris image is presented, and important role ofthe tracking and pointing system is analyzed.2,Aiming at iris localization problem in big scene, eye detection and trackingalgorithm is proposed. The universal background subtraction algorithm theory isanalyzed, and background modeling algorithm for pedestiran movement is designed inbig scene segmentation. The AdaBoost algorithm theory and face color clusteirngalgorithm theory is investigated, and the preparation of the region face detection andface judgment algorithm code is programmed. Gabor filter and gray integral projectionalgorithm theory is analyzed. Hoirzontal or small tilt of the human eye detectionproblem is solved. Kalman filter and MeanShift theories are analyzed and two algorithms for eye tracking are combined.3.Aiming at the problem of iris searching, scheme of automatic tracking controlis proposed. During pedestiran movement characteirstics and the effective area of thesystem，the system precision control and motor requirements is calculated. The paperintroduced the important devices in the system. The motion control algorithm is deeplystudied. The system servo control model is established, and stability of the servosystem is analyzed.4.The implementation process of eye tracking and reflector pointing system isintroduced respectively from two aspects, the perspective of hardware structure andsotfware design. The processing sotfware developed by OpenCV library for eyetracking, and the tracking-pointing effect test of single eye and error analysis iscompleted.