Assessment of the high precision of global least squares stitching over large measuring areas

Abstract

A 3D stitching algorithm was developed to measure wide-range data without reducing lateral resolution. This algorithm can correct X- and Y-axis rotational errors and Z-axis translational errors, as well as minimize the propagation of stitching error using an overall squared stitching method. Furthermore, the coordinate relation of adjacent areas is clearly defined to facilitate proper implementation of the algorithm. The developed algorithm has been applied with simulated and actual measured data to verify that it can stitch various types of data. The stitching accuracy on the common stitching area was evaluated using the Monte-Carlo method as the noise level increased to determine the criteria on measuring conditions. The stitching error was validated to be very small and to occur at the sub-nanometer level.