Abstract: Aiming at the problems of large human reading errors，low measurement efficiency，and poor
adaptability to non-ideal fringes in traditional Newton ＇ s ring interference experiments，an automatic
measurement method integrating YOLOv8 and image processing technology is proposed．The experiment adopts
a strategy combining coarse localization and fine fitting to achieve center localization:first，the YOLOv8 model
is used to quickly locate the center region，followed by sub-pixel-level precise localization through weighted
least squares method．With the center as the reference，grayscale values are sampled radially along multiple
directions，and the edges of dark rings at various levels are identified by combining threshold judgment．Finally，
the radius of curvature of the plano-convex lens is calculated based on the principle of equal-thickness
interference．Using a plano-convex lens with a standard curvature radius of 1 021 mm as the test subject，the
identified result is 1 019．7 mm，with a relative error of less than 0．13%．The total time required for a single
measurement is approximately 4．2 min．
