Research On Coded Structured Light For Three-dimensional Shape Measurement

Three-dimensional shape measurement technique based on structured light has been widely used in machine manufacturing, industry inspection, reverse engineering, object recognition, virtual reality and medical diagnosis because it has many advantages, such as non-contact, fast speed, high precision, low cost, large working distance, wide application, easy to achieve, and so on. Although many advancement has made in one-shot shape acquisition and multi-shot shape acquisition, there are many complex factors during the measurement, for example, surface color, shadow & shading, occlusion, texture, sharp slope, nonstability of the light, interreflection, projector defocusing, subsurface scattering, intensity noise, ambient light, etc. Therefore, it is necessary to improve the measurement precision further. This article focuses on system parameters calibration, color coded structured light, phase measuring profilometry on projector focusing and phase measuring profilemotry on projector focusing, the goal is to improve the calibration precision of system parameters, the decoding precision of color coded structured light and the unwrapping precision of phase measuring profilometry on projector focusing and phase measuring profilemotry on projector focusing when there are many complex factors during the measurement.System parameters calibration is one of the key issues in the structured light system. In this paper, a simple and precise calibration method is proposed. Based on the projective transformation theory, the corresponding relation between camera image and projector pattern is established by projecting color coded pattern which is generated with pseudo-random coding theory. According to the corresponding relation, the image coordinates of the chessboard corner in the projector coordinate system is determined from the image coordinates in the camera coordinate system, so the camera and projector are calibrated simultaneously. Finally, the extrinsic parameters of the structured light system are integrally optimized with the help of pseudo-random coded pattern, and the calibration of structured light system is accomplished. This method is simple and effective, and fit for field calibration. In addition, the precision of the reverse projection error is only 0.3 pixels.The decoding precision of color coded structured light is always low because of the complex factors. In this paper, two decoding methods are proposed for color multi-slit and color stripe, namely hard code-word decoding method and soft code-word decoding method, they includes three steps. First, the location of the feature is extracted with the corresponding detection method. Then, the color of the feature is identified using fuzzy C means clustering algorithm on a color feature with high discriminating power. Finally, the correspondence of the feature between the camera image and projector pattern is determined through the use of multi-step window matching method. The difference between these two decoding methods is that hard code-word decoding method uses the actual code-word to achieve feature matching, while soft code-word decoding method adopts the fuzzy code-word to implement feature matching. Based on the ground truth acquired by space-time analysis, the comparative experiments are conducted. The experiment results show that the color feature only related to the spectral sensitivity of red, green, and blue sensors and the albedo of the surface performs well in color identification, soft code-word decoding method has higher precision than hard code-word decoding method, while hard code-word decoding method owns higher speed than soft code-word decoding method.The unwrapping precision of phase measuring profilometry on projector focusing is affected by the efffecs of interreflection, projector defocusing, subsurface scattering, and so on. In order to improve the unwrapping precision, a micro phase measuring profilometry is proposed, which has lower wrapped phase error than micro phase shifting. In this method, the frequencies of the projected phase-shifted intensity patterns are limited in a narrow high-frequency band. Moreover, a frequency selection rule is developed for micro phase measuring profilometry with the robust Chinese remainder theorm, the wrapped phase standard variance of multi-step phase shifting and the idea of micro phase shifting. After that, the closed-form algorithm on robust Chinese remainder theorm is used in phase unwrapping. The experiment results indicate that the proposed micro phase measuring profilometry not only has high precision, high efficiency and strong robustness, but also can efficiently reduce the effects of interreflection, projector defocusing and subsurface scattering when the frequencies are conformed to the proposed frequency selection rule.Similiarly, the complex factors also affect the unwrapping precision of phase measuring profilometry on projector defocusing. After introducing one-dimensional defocusing projection fringe techniques and two-dimensional defocusing projection fringe techniques, the serpentine scanning method is adopted to improve Floyd-Steinberg error diffusion dithering algorithm, which is aim to generate more ideal grating fringe. Then, all of the defocusing projection fringe techniques are compared respectively through simulation. The simulation results manifest that optimal pulse width modulation method has lower wrapped phase error than squared binary defocusing method, sinusoidal pulse width modulation method and tripolar sinusoidal pulse width modulation method and improved Floyd-Steinberg error diffusion dithering algorithm also owns lower wrapped phase error than Bayer ordered dithering algorithm, traditional Floyd-Steinberg error diffusion dithering algorithm and Stucki error diffusion dithering algorithm. Besides, a hybrid defocusing projection phase measuring profilometry is proposed to reduce the effects caused by the complex factors. Specifically, optimal pulse width modulation method is adopted to yield high-frequency sinusoidal grating fringe and improved Floyd-Steinberg error diffusion dithering algorithm is used to produce low-frequency sinusoidal grating fringe. Finally, the absolute phase is obtained with four-step phase shifting method and multi-frequency phase unwrapping algorithm. The experiment resuts show that the proposed hybrid defocusing projection phase measuring profilometry can reduce the effects caused by the complex factors, as well as has higher unwrapping precision than phase measuring profilometry on optimal pulse width modulation method or improved Floyd-Steinberg error diffusion dithering algorithm.At last, a three-dimensional shape measurement system with structured light technique at the core is developed and the aero-engine blade is choosen as the measured object to achieve the exploratory three dimensional measurement. The experiment results are good.