A novel algorithm to stitch adjacent cloud of points of long cylindrical surfaces

This paper presents an algorithm based on Least Absolute Method to align and stitch multiple adjacent cylindrical clouds of points measured by white light interferometry using conical mirrors. The evaluation of the aligning and stitching algorithm was initially performed by using several numerically simulated clouds of points (COP) of cylindrical surfaces with small shape errors and quite rough surfaces. In order to evaluate the algorithm, each numerically generated COP was split into two parts but always keeping an overlapping area. Numerical translations and rotations were applied in one part to simulate real misalignments. After this, the algorithm was applied to align each adjacent COP pair and to obtain a stitched COP, and the result was compared with the original one. In this way, the performance of the presented algorithm was evaluated and analyzed for several overlapped areas. Excellent results were obtained with an overlapping area of 25% of the total measured length. The differences between the stitched and original cloud of points were always far below the roughness level of the measured surface. A brief description of a modified white light interferometer to measure in cylindrical coordinates as well as early applications of the algorithm in real measurements is also presented.