Research And Development Of 3D Measurement Technique And System For Asphalt Pavement Texture Based On Fiber-optic Interference Fringe Structured Light Projection

Asphalt pavement texture is critical to driving safety,tire road noise and surface drainage performance.The collection of pavement texture morphology information is helpful to deeply analyze and understand the road surface characteristics,so as to ensure protection for the driving quality and safety of drivers and unmanned vehicles,and provide data support for timely pavement maintenance.As a fundamental part of the management and maintenance implementation,there are still some shortcomings in the existing pavement three-dimensional(3D)measurement technique and equipment,especially the lack of high-precision and high-resolution detection equipment with reliable data calibration and controllable measurement process.To this end,this dissertation develops two innovative noncontact active asphalt pavement texture 3D-measurement systems based on advanced fiber-optic interference surface fringe structured light projection technique,combined with monocular and binocular photography geometry,respectively.This study focuses on each step of the device development process,and independently completed the design,fabrication and integration of key hardware in each system,and compiled the supporting optical image and data processing algorithms.The main research contents and results are as follows.(1)The use of fiber-optic dual-aperture interference is proposed to improve the coherent point source generation method in the traditional laser interference technique,which can project high-resolution continuous sinusoidal intensity coded fringe structured light to the measured road surface.Based on the analysis of the principle of dual-fiber interference projection and the properties of interference fringes,the direct or indirect mapping relationship between interferometric phase and pavement texture depth is established by combining the optical path geometry of monocular and binocular measurement systems,respectively.(2)Aiming at the shortcomings of the traditional Fourier transform-based phase demodulation method,two interferometric phase demodulation algorithms based on wavelet ridge detection in different dimensions are given by making full use of the multi-resolution and local analysis properties of continuous wavelets.The strict control equations for 1D and 2D wavelet ridges of interferograms are derived,and it is proved that the phase at the wavelet ridges is the wrapped phase carried by the deformed interference fringes.The phase demodulation performance of the two algorithms is tested and compared by simulation,and the results show that the 2D algorithm is more advantageous.Combining it with the quality guided phase unwrapping algorithm,the phase information of deformed fringes modulated by asphalt pavement texture can be effectively recovered.(3)Interference image preprocessing techniques were investigated.The effective light intensity of the interference fringe is doubled by a single channel grayscale transformation.Two enhancement algorithms are proposed for the grayed fringes,which are adaptive one-dimensional enhancement through modifying intrinsic time-scale decomposition(MITD)and two-dimensional enhancement with full spatial operation,respectively.In which,the MITD in the one-dimensional algorithm is modified by the polarity complementary white noise signal assistance and the separation of fine carrier components can be achieved by the proposed background-carrier signal-noise automatic grouping strategy.The two-dimensional algorithm considers the spatially anisotropic structure of the fringes and improves the traditional coherence enhancing diffusion filtering therein by discrete wavelet threshold shrinkage,thus increasing the adaptability to speckle noise in interferograms.Both simulations and real experiments show the effectiveness of the proposed two enhancement algorithms.The quality of the enhanced interferogram is significantly improved and the fringe phase demodulation results are more robust and accurate.(4)Based on the above principles and the developed algorithm,a monocular+fiber-optic interference projection pavement texture measurement system(MonoTex)is developed.The all-fiber interference subsystem and the image acquisition subsystem in MonoTex are built,including the design and fabrication of a fused tapered 1×2 fiber splitter using the 460-HP single-mode fiber optimized for visible wavelengths to achieve 3dB splitting at 532nm;and the design and fabrication of a flexible four-channel single-mode fiber array to allow accurate determination and adjustment of the core distance while protecting the fiber output end-planes.Through calibration and calculation,the nominal resolution of MonoTex in XY direction is 78.9μm and the Z direction measurement resolution can reach 28μm.And the effectiveness and accuracy of MonoTex for complex asphalt pavement texture measurement is verified by a series of indoor and outdoor tests.(5)On the basis of MonoTex,a more flexible binocular+fiber-optic interference projection pavement texture measurement system(BinoTex)is developed by introducing stereo vision.In particular,the interference subsystem is improved in full polarization maintenance,which completely solves the problem of inconsistent interferometric laser polarization state caused by birefringence of common single-mode fiber and realizes stable and clear projection of interference fringes for a long time;the synchronous hard trigger connection control is realized for binocular industrial cameras,and a high-precision camera calibration and epipolar line rectification process is given by establishing a complete nonlinear perspective projection camera model;aiming at the shortcomings of traditional passive stereo matching based on gray-scale images,an active region phase primitive matching algorithm with dual optimization of efficiency and accuracy is proposed.The experimental results show that the standard deviation of BinoTex for point cloud reconstruction of standard objects(standard plate,standard spherical crown and standard gauge block)can be controlled within 30μm;comparing the real asphalt pavement point cloud reconstructed by BinoTex with the commercial KSCAN20 hand-held scanning point cloud quantitatively,the average absolute errors of all 268363 test point cloud samples is 0.111mm and the standard deviation value is 0.144mm.This dissertation cross-applies the knowledge of fiber optics,machine vision,digital image and signal processing and other disciplines.Two systems(MonoTex and BinoTex)based on fiber-optic interference technique are developed by means of mathematical modeling,theoretical derivation,computer simulation,and indoor and outdoor testing,which provide new ideas and ways for high precision,high resolution,non-contact and full-field measurement of asphalt pavement texture.