Research On 3D Measurement Method Of Complex Surface Based On Projection Grating

The three-dimensional(3D)measurement of complex surface objects has been widely applied in many areas,including reverse engineering,machinery manufacturing,aerospace,electric power,automobile,human body measurement,medical treatment and 3D printing technology etc.such as the engine,the engine case,the reducer and some nuts etc in the agricultural machinery and aeronautics,and most of the key parts are complex in shape and irregular in surface,the level of design and manufacture of these key components is related to national security and core competes.At present,with the development of digital design and manufacturing technology of mechanical parts products,a sharp increase in amount of the overall structure of the integration of various mechanical parts and the requirements for the manufacturing precision of the product are constantly improved,the traditional measurement techniques have been unable to meet the requirements of accurate and rapid detection and measurement for complex surface parts.The current 3D measurement technology based on vision has become a key technical link to improve the efficiency and level of the measurement of complex components.At present,the existing commercialized equipment based on the projection grating is expensive.Therefore,in view of the problem of contour measurement for complex surface objects,the basic principle and key technologies of 3D profile measurement based on projection gratings are studied in depth,a kind of non-contact,fast and high precision complex surface profile measurement is realized by fully utilizing the principle of surface structure light scanning and binocular stereo vision,and the application is verified in the fields of agricultural machinery manufacturing,aeronautical components and human body 3D measurement.The main contents of this paper are as follows:(1)The calibration of camera parameters is one of the key steps in the 3D measurement of structured light.The accuracy of the calibration results directly affects the accuracy of the subsequent 3D reconstruction.To this end,an efficient camera calibration method is proposed.First,a simple,high-precision elliptical center extraction method based on a double cone model is used to extract the elliptical center in the target image,the method directly uses the original gradient information in the boundary neighborhood of an ellipse,and does not need to extract the exact boundary points of each ellipse.Therefore,the calculation process is simplified.In addition,the higher precision is obtained by minimizing the algebraic distance.Secondly,a simple and fast automatic matching algorithm based on random sample consensus algorithm,namely RANSAC(Random Sample Consensus),is proposed,it matches the spatial calibration feature points with their image points at once.The calibration points in the calibrated target are first ordered in a certain order when matching,due to the camera distortion,the extracted feature points in the calibration images do not match the order of feature points in the spatial calibration target,and the corresponding points matching is completed through the RANSAC algorithm.The high precision internal and external parameters of a single camera and between the two cameras are obtained according to the correct matching points and the calibration model of the camera.Finally,experimental results show that the proposed calibration method has high accuracy.(2)Phase calculation includes two processes: phase extraction and phase unwrapping.Phase extraction is two of the key steps in optical 3D measurement based on projection grating.Taking into account the efficiency and precision of complex object measurement,in phase extraction,a phase extraction method based on generalized Morse wavelet is presented,only a single frame stripe image is needed.the Generalized Morse is proposed as the mother wavelet,the continuous wavelet transform is performed on the projected fringe pattern which is deformed by the shape of the object,according to the relationship between the phase of the wavelet transform and the height of the object to determine the 3D profile of the object.And compared with the most popular complex Morlet wavelet as the phase extraction method of the mother wavelet and the Fourier transform profilometry extraction method in detail.Simulation and experimental results show,due to the flexible time frequency local characteristics and exactly analytically properties of the Generalized Morse wavelet,Generalized Morse wavelet as the mother wavelet is suitable,and the effect of measuring the 3D profile is better than that of the popular Morlet wavelet as the mother wavelet and Fourier transform.(3)In fringe projection profilometry,the phase wraps is eliminated completely or the number of phase wraps is greatly reduced in the wrapped phase map,which is helpful for subsequent phase unwrapping.Therefore,an efficient method is proposed to eliminate or greatly reduce the number of phase wraps.Firstly,the sub-pixel spectral peak location is determined by two 1D iterative local Fourier transform.Then the spectrum is shifted to the origin in the spatial domain.After these two steps,it can rapidly and accurately reduce the phase wraps.Finally,the proposed method is tested by simulating fringe images and actual fringe images,and compare with the spectrum is shifted in the frequency domain and the spectrum is shifted in the spatial domain after zero-padding,verified the efficiency of the proposed method.(4)Phase unwrapping is the three of the key steps in optical 3D measurement based on projection grating,especially for the phase unwrapping of complex surface objects,which is one of the main problems of optical 3D measurement.Phase unwrapping is achieved by combining an efficient phase wraps reduction algorithm with the triple frequency heterodyne principle.First,the phase wraps reduction algorithm is applied to handle the wrapped phase images,and then the processed wrapped phase maps are unwrapped by using the triple frequency heterodyne principle.It can better solve the problems of phase unwrapping such as occlusion and space discontinuity.(5)Stereo matching is a prerequisite for subsequent 3D reconstruction,and the accuracy of stereo matching directly affects the effect of 3D reconstruction.Therefore,according to the principle of the polar line constraint and the same phase of the same point,the matching method of the combination of the polar line and the phase is used.First,the effect of distortion is eliminated by correction.Secondly,according to the epipolar constraint,the arbitrary pixel point in the left image,its matching points in the right image must on its polar line,finding the same phase point in the pole line is the matching point,according to the matching point pairs,the three-dimensional coordinates of points are calculated by binocular stereovision principle,the obtained point cloud data are de-noised,streamlined,meshed and optimized.The reconstructed mesh model is obtained,and the 3D entity model of the object is compared with the obtained grid model,and then the error of the point cloud data is analyzed.Based on the above key technologies,a set of 3D measurement system based on projection grating is constructed,which can meet the requirements of industrial application for 3D measurement accuracy and speed.