Research On New Methods Of Phase Extraction In ESPI And FPP And Their Applications On The Dynamic Measurements

Optical measurement technology as an important tool of scientific research,has permeated into many fields of science and engineering.Electronic Speckle Pattern Interferometry(ESPI)and Fringe Projection Profilometry(FPP)are two important kinds of the whole-field,non-contact optical nondestructive measurement technologies.In ESPI and FPP measurements,the physical quantities to be measured are related with the phase information which is encoded in the fringe patterns.The accurate extraction of phase information from the fringe patterns or wrapped phase images is the key to the successful application of the above two optical measurement technologies.The fringe skeleton interpolation method(FSIM)is the most straightforward and widely used approach in phase extraction methods based on single ESPI fringe pattern.In FSIM,the fringe pattern filtering problem and the skeleton extraction problem are the key problems in ESPI phase extraction.Extracting the phase from the single FPP fringe pattern is one of the most widely used FPP phase extraction methods.In the phase extraction method based on single FPP fringe pattern,the background removal problem and the phase unwrapping problem are the key problems in FPP phase extraction.In this paper,we make a lot of in-depth reserachs on the above-mentioned problems.The details are as follows:(1).We propose a general filtering method based on variational image decomposition(VID)for ESPI fringe images with various densities.In our method,a variable density ESPI fringe image is decomposed into low-density fringes,highdensity fringes and noise,and each component is described respectively by the appropriate function space.General density fringe images and high density fringe images can be regarded as the special cases of the variable density fringe images.In addition,we also analyze and compare the performance of the functional spaces used to describe the different parts of fringe images.(2).We propose an intelligent method to achieve fully automated extraction of the fringe skeletons in electronic speckle pattern interferometry(ESPI)based on U-Net fully convolution neural network(FCNN).Once the U-Net full convolutional neural network model is successfully trained,it can be realized the intelligent extraction of the skeleton of multiple ESPI fringe images.By using the trained network,the skeletons of the 1000 frames of the simulated dynamic ESPI fringe patterns are easily extracted.Compared with the existing methods,the proposed method does not need to adjust the relevant parameters in the skeleton extraction process,and it is particularly suitable for processing a large number of multi-frame images.In addition,the influence of noise level and the selection of training samples on skeleton extraction are also discussed.(3).We propose a method for FPP background removal based on the shearlet transform.Background removal is a key step in phase extraction from a single fringe pattern.Shearlet transform is one of the most advanced and excellent methods for the multiscale analysis.We apply the proposed method to the simulated and experimental FPP fringe patterns,and compare it with the most commonly used Fourier transform and empirical mode decomposition methods.Experimental results show that the proposed method is more accurate than the above two methods.(4).We propose two phase unwrapping methods,one of which is the shearlet transfom background binary mask weighted least square phase unwrapping method(STB-WLSPU),the other is the shearlet transfom fringe binary mask weighted least square phase unwrapping method(STF-WLSPU).We apply the two methods to the simulated and experimental FPP phase unwrapping.Compared with five commonly used phase unwrapping methods,the STB-WLSPU is faster in calculation;the STFWLSPU can effectively distinguish the phase continuous regions and the phase discontinuous regions,and avoid the error propagation caused by the phase discontinuous regions while unwrapping.(5).We apply the proposed methods to the ESPI dynamic thermal deformation measurement and the FPP dynamic 3D measurement.We obtain 352 frame skeleton images by using proposed U-net intelligent method and realize the dynamic thermal deformation measurement of alumina ceramics.We apply the proposed FPP background removal method and the phase unwrapping method to the dynamic measurement of human face and hand.The 3D morphologies of facial expressions and hand gestures are restored.