A new Surface roughness measurement method based on image mosaic of template matching algorithm

Abstract

A surface roughness measurement method is proposed to address the current problems: The field of view limit cannot objectively characterize the surface quality of the part, and calculations are complex. A matching template is constructed by overlapping regions, similarity measures are calculated to achieve image matching and splicing, and the double of spliced images is smoothed by a weighted fusion algorithm. The Foreman algorithm is applied to extract the one-sided edge features of the surface roughness, and the least square method is used to fit the contour center line. Finally, the arithmetic mean deviation and the maximum height evaluation model of the roughness are established. The experimental results indicate that the contour length of the splicing of two adjacent frames are increases by 400μm on average. The measurement accuracy is improved by 1.16%.