Research On3D Human Face Imaging Technique And Algorithm Based On Binocular Stereo Vision

Stereo vision is a significant subject in filed of computer vision, reconstructingthree-dimensional(3D) informations is its aim. Although there are measured interfere andconfine of size for pixel of digital images in process of imaging, the precision of regainning3d data is incapable of satisfaction for people, but there are significant applied values onresearching on stereo vision, including independent navigation system of motion robot,avigation and measure for remote sensing, automatzed system for industry and so on.This paper is a branch of science and technology project which is danger detection andsecurity control system for dynamic people stream environment in public of Liaoningprovince education office. The main content in this paper is3D human face imagingtechnique and algorithm based on binocular stereo vision.There are some tasks had to do for implementing3D reconstruction for special humanface:To Calibrate industrial single camera for simulating two cameras(to simulate human twoeyes in this paper)and obtain its inner and outside parameters; To capture head-on and lateralimages of special human face in lab by using camera which is calibrated and image capturecard; To pretreat the images which are captured for enhancing the quality of images; Basingon information in advance, to distill characteristic point for human face, using3D MAX tomake a standard human face model which is read and displayed by OpenGL; To distillcharacteristic for pretreated human face images by using Harris Algorithm which is mended;To get matching point by using least square-method for image matching; To get coordinatevalues in world coordinate, that is3D coordinate values for real human face characteristic byusing mid-point of vertical line in crossover lines to get3D points coordinate; To make wholeand part transform by using human face characteristic for the standard human face model,then make texture mapping.The main productions of this paper are summarized as follow:(1) Completed pretreatment for images which are captured, enhanced the quality of images.(2) Completed the camera calibration and obtained its inner and outside parameters.(3) After3d modeling by Using3DS Max and OpenGL on human face, create a standardhuman face model.(4) After face features extraction, images matching and so on, getting3d reconstructionfor human face, the system can come into vivid3D face succeedly by experiment.