Research On High-Precision Phase Extraction Method For 3D Shape Measurement Based On Grating Image Projection

Three-dimension(3D)measurement technology based on grating image projection has the advantages of high precision,fast and non-contact measurement,aims at 3D digital modeling for measurement object,and has been widely used in many fields,such as 3D printing,digital protection of cultural relics,reverse engineering,industrial detection and intelligent robots and other fields prospects,moreover,it plays an increasingly important role in production and life.The grating projection method relies on the phase information carried by the projected sine fringe image to obtain the 3D shape of the measurement object.However,due to the nonlinear property of the measurement system,the high reflection or non-static state of the measurement object,and the noise during the acquisition of image data,the extracted phase quality from the captured image decreases,which leads to the low-precision reconstruction of the 3D shape.In order to extracting high-precision phase information,the research on phase extraction methods was carried out based on the following aspects.(1)Research on measurement system error compensation method based on grating image projectionDue to the fact that the projectors and the industrial cameras are nonlinear elements,the input and output relation of the measurement system can be expressed by a Gamma nonlinear relationship approximately.The nonlinear property of the measurement system will cause errors on the extracted phase.To suppress the phase error caused by the nonlinearity of the measurement system,a sub-area Hilbert transform(HT)method is proposed.First,the grating image is transformed by the HT and the phase information in HT domain is calculated.Second,according to the absolute phase difference between the HT domain and the time domain,the phase disturbance(PD)area is determined.Then,the phase information in the time domain and the HT domain is averaged directly outside the PD area,at the same time the problem of error compensation is transformed into obtaining a numerical solution to a transcendental equation inside the PD area.Finally,the particle swarm optimization(PSO)algorithm is adopted to solve the numerical solution to the transcendental equation.Simulation and experimental results show that the proposed method can effectively reduce the root mean square(RMS)error of obtained phase and improve the precision of extracted phase.The proposed method does not increase the number of projection images or the costing of the hardware,overcomes the shortcoming of the HT method in spectrum leakage,and inherits the advantages of the HT method.(2)Research on phase extraction method based on high reflection measurement objectBecause of the high reflection,the captured pixels by the camera are saturated in a large area.The saturation pixels will result in a failure to extract phase information,and the 3D shape of the measurement object will not be reconstructed completely according to the phase information.To obtain high-precision phase information,an optimal grating image projection scheme by which the clear images can be acquired is presented.First,according to the mathematical model of imaging,the optimal projection intensity at each pixel in the saturation region is adaptively calculated by projecting three frames of uniform patterns onto the measurement surface.Second,the sequence of gray code pattern is projected,and the mapping relationship between the image plane coordinates and the projection plane coordinates is established.Then,a set of sinusoidal fringe image is projected to verify the reliability of the mapping relationship.Finally,the phase information of the captured clear grating image is extracted using phase-shifting method.The test results show that the presented method can avoid the occurrence of high reflection effectively and capture clear images,and thus can extract high-precision phase to reconstruct 3D shape.The method does not need to change the projection intensity and the exposure time of the unsaturated region,so that the projection images have a higher signal-to-noise ratio.(3)Research on the phase extraction method based on non-static measurement objectWhen the measurement object is in non-stationary state,the extracted phase using phaseshifting method will have a large error,which limits the application of the grating image projection-based method in some scenes.To alleviate the phase errors caused by non-stationary state,the calculation model of phase-shifting is analyzed and deduced,and an improved phase extraction method is proposed.First,the phase calculation model of phase-shifting method is improved.Second,the Morlet complex wavelet is adopted to detect inter-frame phase variation,and the polarity of cosine function for inter-frame phase variation is analyzed.Simulation and experimental results show that the proposed method has the advantages of both the phase-shifting method and the single-frame projection method.When the measurement object is in nonstationary state,the RMS error of the proposed method is the smallest among the comparison methods after 3D reconstruction.It is shown that the phase quality is better using the proposed method.The phase error caused by non-stationary state can be suppressed and the application range of the phase shifting method is extended according the experiment results.(4)Research on denoising method for grating imageIn the grating image projection measurement system,there are noise whose intensity level is unknown in the captured grating image.The existence of noise affects the precision of phase calculation.To suppress the influence of noise on the phase extraction,a filtering method is proposed.First,the relationship between the noise variance and the average energy of grating image is established by modeling the noise with Gaussian distribution,and then a gaussian noise image with zero mean value is added artificially.According to the established relationship,the noise variance of the grating image is estimated.In the second step,according to the estimated noise variance,Gauss filtering is implemented in the phase map.The accuracy and robustness of the noise variance estimation method is tested,and the estimation results show that the proposed method has good performance in estimation accuracy and robustness.The simulation and experiment results show that the RMS phase error of the proposed method is the smallest among all the comparison methods,and the proposed method can effectively suppress the influence of noise on phase.The proposed method provides a theoretical basis for denoising grating image with unknown noise level.