Composite sub-pixel angle point positioning method based on curvature and gray level

Abstract

The invention provides a composite sub-pixel angle point positioning method based on curvature and a gray level and relates to the machine vision precise measurement technology research field. According to the method, firstly image pre-processing on a frame selection interest area is carried out, and burr and oil stain of an original image are eliminated; candidate angle points are extracted through an angle point detection method based on curvature characteristics; multiple dimensioned invariance of an angle point curvature angle and the gray level information in a circular window taking an angle point as a circle center are utilized to eliminate pseudo angle points, and contour end points are considered; the angle point and the contour end points are connected to acquire two straight lines, the two straight lines are taken as reference, edge points of the original image are screened, a point set of the two straight lines of to-be-detected angle points is acquired, the least square method is utilized for fitting to acquire two straight lines, and the intersection point is the angle point. The method is advantaged in that a problem of reduction of angle point detection reliability and accuracy caused by attachment existing in shaft workpieces is solved, the acquired angle point is closer to an actual angle point, and precision reaches a sub-pixel level.

Description

technical field [0001] The invention relates to the research field of machine vision precision measurement technology, and specifically refers to a corner point sub-pixel positioning method based on the combination of curvature and gray scale. Background technique [0002] Shaft workpieces are very common and important rotating workpieces in the industry. The comprehensive shape error accuracy not only directly affects the motion performance and service life of the machine, but also has an important impact on reducing energy consumption and environmental pollution. In industrial applications, manual measurement is usually used to measure the error of shaft workpieces, and its measurement accuracy is not high. Meet the requirements of large-scale online measurement. The measurement method based on machine vision has the advantages of non-contact, fast, accurate, and easy automation. Among them, the test system composed of a line array camera, a parallel backlight and a moti...

AI Technical Summary

Problems solved by technology

Based on the above interference, if the corner points of the original image are directly extracted, many false corner points will be obtained, which will affect the reliability of corner point extraction

Images