Finding optimal correspondence sets for large digital metrology point clouds using anisotropic diffusion analogy

ABSTRACT

The inspection process in today’s digital production cycles is acyber-physical complex including the physical collection of high-quality metrology data and extensive computational tasks to generate valuable information following, intermittently, or online with the manufacturing tasks, to support quality control or quality improvement of products, or production system’s health monitoring and prognostics. In this paper, amethodology to estimate the geometric deviation zones of large coordinate metrology point clouds is presented based on afinite difference approach relying on the anisotropic diffusion analogy. Instead of heat, the normal distances between the ideal model and the measured points are considered as the independent variables in the diffusion process with respect to the time and location. The location term is discretized by finite differences and the temporal term is estimated by the forward Euler method. In order to eliminate saturation in the diffusion process, the diffusion coefficient dynamically changes in each iteration based on the variations in the standard deviation of the Euclidean distances. The developed methodology is fully implemented and avariety of validation tests and experimental studies are conducted. According to the results of the virtual tests and the experimental studies, the developed methodology reduces on average about 60% of the number of measured points and consequently the computation time, even for highly complex surfaces with considerable inherent manufacturing errors, without imperiling the accuracy of the deviation zone estimation.

KEYWORDS:

• Deviation zone estimation
• coordinate metrology
• anisotropic deviations diffusion
• forward euler
• finite difference
• digital metrology
• point clouds
• data reduction

Disclosure statement

No potential conflict of interest was reported by the author(s).