Research On Zoom High Precision Binocular Stereo Vision Measurement Technology

It is necessary to reconstruct the 3D shape of object obtain the local feature accurately and robustly.Comparing with the fixed field measurement of the traditional fixed focus measuring system,zoom measurement system can not only reconstruct 3D shape of the object but also adjust the size measurement field to further improve the accuracy.Therefore,this paper establishes the stereo vision system based on the zoom cameras and researches deep into the binocular zoom measurement system model,image feature recognition and matching in different scales,calibration of binocular zoom system based on the feature points,and point cloud reconstruction and splicing.The main contents of this paper are as follows:(1)The mathematical model of binocular zoom stereo vision measurement system is studied,and the conversion relation between each coordinate system and the epipolar constraint of binocular measurement system are deduced.(2)This paper focus on matching algorithm based on the scale invariant feature transform feature points,analyses the homography matrix relationship between images,and proposes feature points matching method based on image contour,and improve the feature points in different scales.(3)This paper focuses on the Zhang's plane calibration method of fixed focus camera,and selects a focal length of zoom measurement system as the initial state of the system.when changing the focal length,it proposes a calibration method of binocular zoom system which is step calibration based on feature points and Zhang's plane calibration method,and then it improves the flexibility of zoom system calibration.(4)This paper analyzes the phase shifting method for 3D reconstruction.Under the different scales of the zoom measurement system,reconstructing point cloud of object based on unwrapped phase and splicing point cloud.