Theory And Methods Study On Fringe Analysis For High Resolution Optical Phase Three-Dimensional Measurement

Optical phase 3D measurement technology is widely used in industrial,entertainment,medical and other fields of three-dimensional surface topography measurement and internal tomography detection due to the advantages of high precision,high resolution and non-contact.In the research of phase measurement,fringe is an important link between intensity and phase,so the theory and method of fringe analysis have become the research focus in the field of optical measurement.The fringe analysis can be divided into single-frequency fringe analysis and mixing fringes analysis according to the application scene.Single-frequency fringe analysis is used for surface phase measurement represented by the Fringe Projection Profilometry(FPP),which enables high precision 3D morphology reconstruction for complex objects.In contrast,mixing fringes analysis is used in tomographic phase measurement represented by the Wavenumber Scanning Interferometry(WSI),which can be used to measure the internal phase field of composite materials.Although the above two phase measurement technologies based on fringe analysis have shown superior performance and great development prospects in surface measurement and tomography measurement respectively,there are still some problems that need to be solved in the practical applications:1)In the online 3D inspection of industrial pipelines,the traditional FPP system always ignores the phase error caused by the deviation of the reference plane.2)In industrial measurement,objects with discontinuous surfaces and different reflectivity is common.These objects will cause both shadow regions and light intensity saturation in fringe images,which will seriously affect the quality of phase recovery.3)Speckle noise is an inevitable interference factor in wavenumber scanning measurement,which not only affects the accuracy of phase extraction,but also causes phase unwrapping errors.4)In tomographic measurement,although the increase of the scan bandwidth can improve the depth resolution,it will also increase the wave number nonlinearity effect,resulting in an increase in the phase extraction error of the mixing fringe signal.To solve the above problems,this article analyzes and studies the principles of two fringe analysis methods,and builds two phase measurement systems based on fringe analysis.Further,this article proposes the following solutions:1)An enhanced FPP system based on reference phase reconstruction is proposed.Compared with the traditional measurement system,this system can realize the real-time update and correction of the reference phase.Then,a phase reconstruction algorithm based on Total Variation Minimization(TVM)is proposed for high-precision reconstruction of the reference phase and smooth noise reduction,which realizes the synchronous acquisition of the reference phase and the object one.Therefore,the measurement results will not affected by the moves or shifts in the reference plane of the FPP system.Experimental results illustrate that the proposed method shows competitive performance with traditional FPP in accuracy and more robustness in complex industrial scenes.2)A high dynamic range FPP system based on backlighting technique and multi-exposure technique is proposed.The system utilizes the characteristics of backlighting to realize the shadow compensation of discontinuous surfaces.After shadow compensation,the relatively time-consuming temporal phase unwrapping(TPU)can no longer be used in traditional single-frequency fringe analysis,which provides an opportunity for the application of simple and efficient spatial phase unwrapping(SPU).On the other hand,the high dynamic range imaging of multi-exposure can reduce the influence of large range of reflectivity variation in multi-color objects and further realize fast and robust three-dimensional measurement3)According to the different mechanisms of Fourier Transform(FT)and Least Squares Algorithm(LSA)to determine the speckle noise points in the phase extraction in the mixing fringes,a phase noise reduction algorithm called Syncretic Fourier transform and Least squares Algorithm is established(SFLA).The algorithm uses FT and LSA to locate the noise points in the speckle phase respectively,and fuses the two noise point map into a phase with the least noise points.Then,a path-following method is established to unwrap the speckle phase.The experiment verifies the performance of the proposed method on speckle phase noise reduction4)The phase estimation problem in mixing fringes analysis is casted into the blind signal separation problem,and a new mixing signal parameter estimation algorithm based on blind signal separation is proposed to realize deep super-resolution.The proposed method first uses the relationship between the adjacent three frames of sine signals in the wavenumber sequence to establish an equation based on the blind signal separation theory,and then employs the Lagrangian multiplier method to estimate the amplitude,frequency and phase of the mixing fringes iteratively.Finally,the proposed method realizes a measurement performance that broke through the theoretical depth resolution under a narrow wavenumber scanning range.Experimental results prove that the proposed method can accurately separate the multi-surface phase distribution in the optical source of narrow bandwidth that FT cannot resolve.