Research On Mirror Reconstruction Technology Based On Binocular Vision And Phase Measurement Deflectometry

With the development of science technology and in-depth industrialization,the produce and application of high-reflective mirrors and mirror-like industrial products is also increasing.At the same time,higher requirements of processing and measurement accuracy of mirror and mirror-like industrial products is increasing.The existing optical measurement methods based on two-dimensional image processing use industrial cameras to get the photograph of products,which can be converted into two-dimensional pixel information by computer image processing,but they cannot obtain the depth information and three-dimensional information.Therefore,more and more researchers begin to study the high efficient and accurate 3-D topography measurement technology to suit the increasing need of mirror or mirror-like products application.They have got many research results.The widely used binocular stereo vision method has the advantages of simple structure,simple algorithm and fast measuring speed,but there exists the problem that difficult to match due to the lack of characteristic points in the case of mirror-like products measurement.In order to solve the problem when measuring the mirror products,the phase measurement deflectometry(PMD)came into being.The principle of PMD uses the relationship between the fringe phase and the gradient to get three-dimensional information of the products,which is used to measure the surface of mirror and highly reflective products.But this method exits the problem of height polysemy.Those two methods also have weakness and advantages.This paper studies a measurement system based on binocular vision and phase measurement deflectometry by combining the advantages of the two methods to measure the surface of mirror and mirror-like products.The aim of that is obtaining 3-D model of the surface.The main research contents as follows :First of all,this paper introduces the principle of measurement based on binocular vision,then the principle of phase measurement deflectometry is shown.After that,this paper introduces the phase measurement deflectometry based on binocular vision by combining those two methods.The key technologies of the combined methods are introduced and analyzed.Secondly,this paper introduces the principle of system calibration technology in detail.This technology is achieved include binocular vision calibration and display screen calibration.The common coordinate Systems in camera calibration is introduced in detail.Zhang's calibration method is used to get the parameters of cameras.The mirror is used to project the mirror image of the display screen calibration plate into the camera's field of view.This method can solve the problem that same-side cameras cannot directly capture images of calibration plate,thereby determining the spatial position relationship between the camera coordinate system and the display screen coordinate system.Moreover,the paper introduces the fringe phase extraction and unwrapping technology,the multistep phase shift method to extract the deformation grating fringe phase is introduced in detail.Various phase unwrapping techniques are introduced and analyzed.Three-frequency heterodyne phase unwrapping method can effectively remove the influence of phase error.Compared with other phase unwrapping algorithms,the calculation of three-frequency heterodyne phase unwrapping method is simpler,with higher efficiency and lower time complexity.After that,the principle of binocular vision correlation is used to determine the surface normal vector,and then the gradient data is derived.Various gradient integration algorithms are introduced.The advantages and disadvantages of various algorithms are analyzed.the Southwell integral reconstruction algorithm is selected to process the derived gradient data.Finally,a measurement system for experiment based on binocular vision and phase measurement deflectometry is built.In this paper,the smartphone is taken as the object of measurement,the surface of its front screen is measured and the three-dimensional surface model of screen is reconstructed to verify the feasibility of the algorithm.