Please use this identifier to cite or link to this item:

An absolute 3D measurement method using a 2D pattern based on digital moiré

Authors Liao, Jiarui
Issue Date 2012
Summary With a remarkable evolution in the development of digital cameras, non-contact 3D measurement using computer vision has been rapidly developing in the past few decades. Excellent and well-accepted techniques include digital moiré and stereo vision. An obvious drawback of digital moiré is that it is not applicable to the measurement of significant discontinuities, which is common in this area of research. And in the process of measurement, error propagation is inevitable, especially under the condition of missing points. The reason for this shortcoming lies in its relative coding scheme. For stereo vision, it is based on triangulation which leads to an absolute measurement. However, correspondence searching and resolution limitation of measurement of lacking features are two of the major unsolved problems of stereo vision systems. Inspired by these two well-developed techniques, an absolute measuring method based on structured light is developed in this research, requiring only a camera and a projector. The philosophy behind the coding scheme is a hybrid of relativity to the center and absoluteness, or independence on every single point. The mathematical model of the system is described and theorems that relate to the guidance of designing such a system are introduced. Two important factors determine the accuracy of the 3D measurement: correspondence matching and calibration of the camera and the projector’ parameters. For the correspondence matching, an image processing method is developed. Image subtraction, edge detection, grid permutation and establishment of sub-coordinate-systems are included in the algorithm. Aiming at the discontinuity measurement, discontinuous borders are marked out first by using a Gabor filter first. Epipolar geometry is then utilized in the process of searching out each corresponding points accurately on the image plane. Sub-pixel accuracy for correspondence matching can be achieved in this way. The second factor that affects the accuracy of the measurement system is concerned with the calibration of the whole system. A convenient calibration method for structured light systems is developed in this research. This calibration method significantly simplifies the calibration procedures, and experimental results are presented for the verification of this calibration method. And experiments verify accuracy of the whole measurement system. And its advantages over digital moiré and stereo vision are presented and verified by experiments. However, this calibration method for structured light system is applicable to the existing methods that are based on the least squares minimization effort. The assumptions made for the calibration system are not suitable to the nonlinear stereo vision system. Moreover, the 2D planar pattern used in existing calibration methods cannot provide 3D information in the 3D space. To move it among different views will inadvertently introduce additional mechanical errors. An iterative calibration method is developed in order to solve the existing problems in the state-of-the-art calibration methods for stereo vision systems. It is the first time that an iterative calibration method is developed to solve the calibration of a two-camera system’s calibration with iterations in the 3D space. Derivations are described for this iterative calibration algorithm using feedback control theory. Furthermore, the calibration target is designed and manufactured under specifications defined by the task has been created for. Experiments show the developed iterative calibration method based on feedback control can achieve convergent parameters. With the calibrated parameters, 3D measurement of the calibration target verifies the correctness and accuracy of this iterative calibration method. Hence, the contributions of this work are significant. An absolute 3D measurement system based on a 2D pattern using digital moiré has been developed. At the same time, a novel calibration method for such a system has also been proposed and verified in this work. For investigation of calibration method on stereo vision system, an iterative method based on feedback control is developed. The presentation of such a complete system is a significant step forward in the field of 3D measurement systems.
Note Thesis (Ph.D.)--Hong Kong University of Science and Technology, 2012
Language English
Format Thesis
Files in this item:
File Description Size Format
th_redirect.html 341 B HTML