Research On Phase Extraction Algorithm For 3-D Shape Measurement

With the development of the information technology,three-dimensional(3D)information of objects played a great role in many fields,such as,quality control,reverse engineering,target detection,product inspection and digital model generation.The development and research of 3D shape measurement algorithms has increasingly become the focus of experts in various fields.Among them,Fringe-projection profilometry is widely used because of their advantages of non-contact operation,high resolution,full field measurement,high data processing efficiency,simple equipment requirements and high reliability.However,as the requirements of various industries gradually increase,the accuracy and speed of measurement need continuous breakthroughs.The accuracy of the 3D shape measurement depends on the quality of the extracted phase.Therefore,the focus is how to extract the phase quickly and accurately.Besides,most phase extraction algoritihms can only obtain the phase that is wrapped within(?,?),which needs to be unwrapped.Therefore,the development of fast and reliable phase extraction algorithms is still a difficult point in the research of 3D shape measurement.This paper studies the problem of phase unwrapping from both speed and accuracy.The main research contents are as follows:(1)To apply the Fringe-projection profilometry to high dynamic range 3D shape measurement,Jiang combined the regular and reverse fringe pattens to compensate pixel saturation by complementary grating stripes.To solve the problem in Jiang's method,a generalized phase-shifting method is introduced to propose an enhanced high-dynamic range 3D measurement method.For each pixel position,all unsaturated fringe data are valid at the corresponding positions in multiple arbitrary phase-shifting fringe patterns.Theoretical and experimental results show that the proposed method significantly improves the accuracy of Jiang's method.(2)Based on the analysis of the Fourier transform method,we found that,it is difficult to separate the signal lobes by band-pass filtering in Yun's dual-frequency Fourier transform method since the positive and negative low-frequency components are close to each other.To solve this problem,a dual-frequency phase demodulation algorithm based on spatial-temporal fringe method is proposed,which improves the accuracy of phase extraction of 3D shape measurement.(3)The time domain phase unwrapping algorithms are studied.For the dual-frequency(or dual-wavelength)time domain unwrapping algorithms,noise is large,phase shift is fixed,and calculation volume is large.The general phase-shifting method is used to break the limitation of the phase shifts.The composite fringe projection method is introduced to reduce the amount of patterns.The period of the low-frequency fringe pattern is increased and the frequency ratio of high and low frequency is reduced,so that the error is reduced.In the interferometry,an iterative two-step time domain phase unwrapping method is introduced to improve the sensitivity and accuracy.(4)To solve the problem of Fourier transform that the peak frequency shift operations in the frequency domain is limited to the whole pixel,we propose a wraps reduction method based on iterative discrete local Fourier transform.It can accurately position the signal spectrum,so that the phase wraps can be eliminated accurately.(5)To solve the problem in Perciante's method that it is only applicable to continuous surfaces,we proposed a spatial domain phase unwrapping algorithm based on the weighted least squares method.The binary quality map is used as a weighting factor.The proposed method can effectively detect and process the noise,shears,shadows and inconsistency,so that the accuracy is improved.(6)Above algorithms are achieved using MATLAB R2014a.The experimental results demonstrate the validity and reliability of various phase extraction algorithms proposed in this paper.