Research On Fast Digital Projection Based Three-dimensional Fringe Pattern Profilometry Techniques

Digital fringe projection profilometry (DFPP) as an effective technology for non-contact3D shape measurement has attracted extensive research in recent years due to many potential applications in enormous fields including industrial manufacturing, medical engineering, homeland security, archeology and entertainment. The idea of DFPP is to employ a light pattern with a special designed structure to probe the object surface of interest, and extract the3D shape of the surface from the reflected light pattern. In early years, the light patterns are generated by optic and mechanical systems. With the advanced digital projection technology in recent years, digital video projectors have become an attractive option, leading to a simple and cost effective implementation of the structured light technology, namely DFPP. Based on the low-cost optical and electronic hardware, DFPP technology could acquire and process large amounts of3-D data at a high speed and accuracy.However, existing DFPP technology is still considered to be premature, as the performance in terms of measurement accuracy and speed is still not good enough for many practical applications. We focus on the research of intensity ratio calculation, intensity ratio unwrapping and intensity ratio-to-height conversion in DFPP based3D shape measurement technology. Besides, a low-cost, high accuracy and speed3D profile measurement system will be set up.In DFPP technology, the major techniques are phase detection based algorithms and intensity ratio detection based algorithms. We introduce some algorithms based on the measurement procedure of phase and intensity ratio calculation, phase and intensity ratio unwrapping, and phase and intensity to height conversion. Besides, we also compare these two algorithms. The intensity ratio algorithms have the advantage of high processing speed but more or less suffer from the ambient light or object surface reflectivity. In the phase detection based algorithms, phase shifting profilometry (PSP) is considered as the most effective method, which is free of influerence of ambient light and object surface reflectivity. Compared to intensity ratio methods, PSP has higher accuracy, but the computation of an arctangent function to extract the phase in PSP brings extensive computational cost. Both these two kinds of methods do not solve the problem casued by nonlinear distortion, which is considered as the major measurement error source.In order to design a fast and accurate algorithm to reconstruct the3D object shape, we present a multiple-step triangular-pattern spatial shift estimation method. The proposed approach is based on intensity ratio from triangular patterns and spatial shift estimation method (SSE). Compared to exsiting triangular patterns based intensity ratio methods, the proposed method do not suffer from nonlinear distortion, so it has higher measurement accuracy. Triangular fringe patterns are very efficient in terms of computational burden when using SSE algorithm. Compared to PSP algorithm, the proposed method has higher processing speed, and only needs at least two image patterns to reconstruct the3D object shape, while PSP needs at least3. Experiments show that the measurement error is reduced by this method in contrast to the triangular-pattern phase shift method.In the exsiting triangular patterns based intensity ratio method, the intensity ratio calculation is influerenced by background illumination and surface reflectivity. Therefore, a novel intensity ratio calculation approach based on multiple triangular patterns is also proposed. All intensity ratio based techniques are advantageous by reduced computational burden, but they are sensitive to either surface reflectivity or background illumination. Besides, those methods require the neighborhood pixel properties to calculate the intensity ratio and still suffer from nonlinear distortion. Due to the point-to-point operation, this new intensity ratio method does not suffer from the influence of the surrounding points, and it is also less sensitive to surface reflectivity and background illumination. By combining the SSE approach, the3D object shape measurement, which is based on intensity ratio, will not be influenced by the system nonlinear distortion. Besides, the processing speed of fitting the intensity ratio by polynomial is the same as the traditional intensity ratio calculation method based on triangular patterns. Experiments show that the measurement accuracy can be increased by employing this method compared with the triangular-pattern spatial shift estimation method, and both these two algorithms have the same speed in the process of intensity ratio curves fitting by polynomial.Based on the research of DFPP techniques, we have developed a DFPP system for3D object shape measurement. This system includes the basic equipments such as digital projector, CCD camera and high performance PC. Besides, optical platform, electrical panning control console, electrical rotating control console, and step motors are also included to achieve high performance in this system. The system software can achieve the functions of projecting various types of patterns, acquiring pattern images, processing3D data, and3D data visualization.Finally, we conclude the thesis and look forward to future works.