【作者】 耿飞；

【导师】 吴成东； 王晓哲；

【作者基本信息】 东北大学 ， 模式识别与智能系统， 2008， 硕士

【摘要】 三维重建是计算机视觉的主要目的和计算机辅助设计的核心问题。在各种建模和重建技术中,结构光技术以其简单、非接触、测量范围大、精度高的特点,获得了广泛的应用。本文研究了一种基于彩色条纹结构光的三维物体重建方法。主要工作包括彩色条纹的编码、颜色区分算法、边缘提取和解码算法、摄像机和投影仪标定方法、三维信息计算和三维重建方法。首先,研究了彩色条纹的编码。提出了利用RGB颜色立方体八个顶点值对彩色条纹进行编码的方法,所采用的特定编码顺序保证了各条纹编码的唯一性,有利于后续的解码处理。设计了一种新的边缘提取方法,对彩色图像中R、G、B三个通道分别提取边缘,再将结果进行合并处理获得图像的边缘信息。实验效果表明此方法可以得到较完整准确的边缘。第二,研究了将颜色信息从RGB空间转换到HSI空间中,实现对颜色区分的算法,以去除亮度不均的影响。提出了一种结合阈值和HSI空间的分色算法,利用阈值完成对于黑色和白色的识别,在HSI空间对彩色信息进行区分,实现对于彩色编码中使用的八种颜色的区分。提出了一种基于条纹颜色的解码算法,解决了物体表面不连续时的解码问题。实验结果表明,分色和解码算法可以实现较好的分色效果和准确的解码。第三,研究了使用标准板对摄像机进行标定的方法。利用标准板上标志点中心的空间坐标和图像坐标获得最优的摄像机投影矩阵。基于摄像机的投影矩阵和标准板平面方程完成了投影仪的标定。最后,研究了通过摄像机和投影仪的标定参数计算物体表面三维信息的方法。研究样条插值运算的算法并利用样条插值实现对物体表面的三维重建。通过实验证明了文中提出方法的有效性与先进性。更多还原

【Abstract】 Three-dimensional reconstruction is the main purpose of machine vision and key issue in computer-aided design. Among the various modeling and reconstruction technologies, structured light is being widely used because of its advantages such as simplicity, no touch, wide detection range and high accuracy.The research of three-dimensional reconstruction based on colorful stripe structured light is addressed in this thesis. The main work of the thesis includes coding and decoding strategy of the colorful stripe, color identification strategy, edge detection method, camera and projector calibration, three-dimensional position calculation and reconstruction.Firstly, a coding strategy for colorful stripe is proposed in the thesis. To meet the uniqueness requirement of code word, which can simplify the decoding process, a new type of projection pattern based on the colors of the eight corners on the RGB cube is proposed in the thesis. In order to detect the complete and accurate edges in the colorful image, an edge detection method is proposed which firstly extracts the edges in R, G, B channels separately and then combines these three results together. The credibility of this edge detection method has been shown by the experimental results.Secondly, in order to minimize the interference effect of lightness, the color data is transformed from RGB space to HSI space. A color identification algorithm combining channel threshold and HSI space is proposed, which identifies black and white colors by threshold and the other six colors in HSI space to accomplish the color identification of the eight colors used in coding. A decoding strategy using the stripe color is addressed to overcome the discontinuity problem. The results indicate that the proposed methods are effective.Thirdly, in the calibration process the world and image coordinates of the marks’ centers on the calibration plate are applied to optimize the camera projection matrix. The calibration of the projector is accomplished on the basis of camera projection matrix and plane equation of the calibration plate.Lastly, the three-dimensional position calculation is carried out by using the parameters of the camera and the projector. The three-dimensional reconstruction is implemented based on spline interpolation. The effectiveness and advantages of the proposed method in the thesis are proved by the experimental results.更多还原