反求工程中消除CT切片轮廓奇异点研究

Study of Removing Singular Pixels of CT Slice Contour in Reverse Engineering

在基于工业CT的反求工程中,轮廓跟踪是其重要的步骤之一。本文分析了边缘检测后可能存在的断点、毛刺、双像素宽边界等奇异点,给出了8邻域和5邻域相结合的跟踪算法。该算法能够有效地消除奇异点,实现了连续单像素链序列表示的封闭轮廓曲线。通过对某车辆发动机的实验,表明采用该算法能够准确表达实物的边界,具有较好的实用性。

The contour tracking is one of the important steps in reverse engineering based on industrial computed tomography. In this paper, the singular pixels of CT slice contour after edge detection, such as discontinuity, redundant, double pixels, are analyzed and the contour tracking algorithm combined eight neighborhood calculation with five neighborhood is presented, which removes the singular pixels effectively and gives a close contour with single-pixel connection. It is shown by the experiments of a vehicle engine that the contour can be obtained exactly and the method is practical.