Optimal Solution For Train Wheel 3D Measurement On Abrupt Surface Based On Fourier Transform Profilometry

To achieve a thorough 3 Demensional(3D)measurement of train wheel surface,and get the precise data of wheel profile,wheel tread defect like scratch and peeling,this paper do research works on Fourier Transform Profilometry(FTP)to find the optimal solution for train wheel 3D measurement,especial for those part with abrupt edge.Firstly,FTP key techniques are theoretically analysed,such as light path structure,stipe projection and distortion correction,frequency filting and phase abstracting,phase unwrapping,3 demensinal calibration,3D data matching.Secondly,optimal solutions in each step are studied to adapt to the CRH train wheel surface,which features in a large scale in whole while small sized defect on wheel tread,and there are abrupt edges on wheel flange and tread defect.Finally,optimal parameters and adaptive algorithms are validated byMatlab simulation and real CRH wheel experiment testing platform.Those optimal parameters and adaptive algorithms are listed as bellow:1.To obtain an optimal 3D reconstruction performance,different period of sinusoidal strips and different projecting derection of structured light illumination is studied.The conclusion is drawn that 3D measurement of train wheel can be achieved while the period of sinusoidal strips varies from 1/16 to 1/8(Strip/Pixel).2.On the aspect of image pre-processing,correction algorithms for keystone distortion and rotation distortion of sinusoidal strips projection are studied,which effectively reduces the influence of distorted strip and improves the 3D reconstruction accuracy.3.FTP and the improved methods of WTP(Wavelet Trasform Profilometry)and STP(S-Trasform Profilometry)are compared studied,to select the fittest method for phase ing from abrupt area of wheel surface.The optimal frequency filter for FTP is orthogonal eclipse filter,while the optimal mother wavelet for signal analysis is cmor0.7-1(one of Morlet wavelet),and the optimal regional frequency filting window is flat Gauss filter.It is found based on real CRH wheel experiment that three methods can correctly the warped phase for train wheel.But WTP and STP are better than FTP in phase ion for abrupt edge area,while WTP is the optimal method for 3D measurement of abrupt area because of its higher accuracy and adaptability.4.As phase unwrapping for concerned,unwrapping algorithms related to and unrelated to path are compared studied.Relative reliability-guided phase unwrapping algorithm based on region partition is the optimized algorithm for train wheel 3D measurement,which can solve the problem of ’stretch’ for phase unwrapping in abrupt area.5.Based on ’Chess Board’,camera is calibrated by Least Square Fit method.Based on 7 cones,the processing of phase-height calibration is fulfilled and simplified,and the calibration is speed up,but the accuracy of calibration should be improved.6.For 3D data matching,random filtering for noise reduction and uniform grid method are adopted to make 3D data sparse.Rough matching is conducted by analyzing key component.Refining matching is achieved by ’k-d tree’ searching algorithm,based on which,the whole wheel 3D surface in 360 degree is fulfilled by 25 frame of distortion stripe image.By combined use of those optimal parameters and adaptive algorithms in key steps of FTP,a rapid and precise inspection is achieved not only for the whole wheel 3D surface in large scale,but also for the tread defects in small size.So,this paper provides a new solution to achive a thorough inspection of wheel surface with high efficiency and accuracy,which surpasses the limitation of traditional inspection methods.The conclusion is draw that,wavelet transform analysis is the best way to abstract phase from abrupt edge of wheel surface.