Structured Light Based High Dynamic Range Shiny Surface 3D Shape Measurement Technique And System Development

Structured light three-dimensional (3D) measurement technique has been used in a broad range of applications including quality control/inspection, reverse engineering(digitization of complex, free-form surfaces), object recognition, documentation of cultural artifacts,medicine and virtual reality. The 3D shape of a workpiece can be obtained precisely by this technique, and thus it provides an effective guidance for the advanced manufacturing, automatic assembly, surface inspection and so on. On the other hand, the 3D point cloud of the workpiece can be further compared with the computer aided design (CAD) model or the existing 3D data to detect the workpiece deformation,and to provide a more complete and easier way for quality control, stress analysis, and collision test. However, the use of this technique still involves difficulties. For example,it requires ambient light in a limited range when measuring, and the measured object having a diffuse or near diffuse reflectance and low reflectivity variations from point to point. In practice,a large number of metallic workpieces are needed to be milled in industrial applications. After milling, the workpiece surface will become very bright. If use the conventional structured light method to directly measure this shiny surface, the light reflected become so strong that image saturation occurs. Thus, the fringe patterns modulated by the highlight cannot be correctly decoded, and thus the corresponding regions on the surface are not measurable. On the other hand, due to the limited dynamic range of the camera, the region with low reflectivity will cause the fringes to be dark,leading to greatly reduce the measurement accuracy. In order to solve these problems, we deeply studied structured light pattern coding strategy, phase error compensation, system calibration and high dynamic range (HDR) image acquisition, and propose a novel adaptive HDR structured light 3D shape measurement technique including line-shifting algorithm, feature-mapping-based system calibration algorithm, adaptive HDR structured light 3D shape measurement algorithm, and phase error compensation algorithm based on smoothing spline fitting. It can effectively solve the problem of data loss in the 3D shape measurement of shiny surfaces, and provide an effective solution for the HDR 3D shape measurement, especially for the 3D measurement of complex machining workpieces. The research contents and main conclusions of this thesis are as follows:(1) The 3D shape measurement techniques for shiny surface are reviewed,summarized, compared and analyzed. Some relevant problems in this field are discussed and the research contents are confirmed.(2) To overcome image saturation, interreflection, and high sensitivity to noise in the measurement of shiny surface by the widely used phase-shifting algorithm, we propose a fast, robust and high spatial resolution line-shifting algorithm based on the theory of spatial coding, temporal coding and phase-shifting algorithm. In order to make the pattern more reliable than the sinusoidal fringe pattern, generate normal and inverse stripe patterns by the Gray code, and shift the stripe patterns like phase shifting. In the decoding procedure, the nonlinear contour of the stripes is interpolated linearly. By solving the intersection to obtain the sub-pixel edge, this algorithm can achieve a higher spatial resolution.(3) System calibration is a complex and time-consuming process. The reference-plane-based system calibration algorithm has some problems, such as too high constraint, not high calibration accuracy, poor operability, and requirement of special customized checkerboard. Aiming to solve these problems, we propose a feature-mapping-based system calibration algorithm. This algorithm treats the projector as a camera, and maps the coordinates of features in the captured checkerboard images to the projector coordinate system by establishing the correspondence between camera pixels and projector pixels. It converts the projector parameters calibration into the mature camera calibration, and thus simplifies the structured light system calibration to a classic stereo vision system calibration.(4) Aiming at the problem in 3D shape measurement of shiny surface, i.e., lack of a complete and adaptive method according to the measured object, we propose an adaptive HDR structured light 3D shape measurement algorithm. By establishing a novel mathematical model for pattern imaging on shiny surface, we analyzed the effects of surface reflectivity, interreflections, and ambient light illumination. As a result, by adaptively adjusting the pixel-wise intensity of the projected stripe patterns, this algorithm can overcome the above effects caused by high light and dark, so as to obtain a clear stripe image, and restore 3D shape of the measured object.(5) During phase calculation, for the phase error caused by the non-sinusoidal attribute of the captured fringe image, we established a mathematical model for the system nonlinear effect, analyzed the system nonlinear response and phase space distribution characteristics, and propose a phase error compensation algorithm based on smoothing spline fitting. The phase error look-up table is created from the stripe images of the planar calibration board, and used to compensate the calculated phase in the subsequent measurement. For the residual phase error after compensation, the smoothing spline fitting method is used to further smooth the phase.(6) Based on the investigations mentioned above, we designed and developed a structured light 3D measurement system for shiny surface. By using the Qt application development framework and the OpenCV computer vision library, we implemented the system function modules. Finally, the effectiveness of the measurement system is verified by several applications.