An improved measurement method of size of mechanical parts based on monocular vision

Abstract

During the monocular measurement of multiple mechanical parts with non-negligible thickness at the same time, the physical dimensions of the pixels on the upper and lower surface edges of the mechanical parts in the image taken by the camera are different, resulting in the deviation of the actual measurement results of the mechanical parts In this paper, a measurement method based on unit pixel size compensation is proposed. This method compensates the measurement data by the distance between the upper and lower surfaces of the mechanical parts and the camera, thus reducing or even eliminating the measurement error caused by the thickness of mechanical parts. The simulation results show that, under the same measuring platform, for the mechanical parts of different sizes and thicknesses, this method can achieve a significant improvement in measurement accuracy.

A Cost-Effective Method for Automatically Measuring Mechanical Parts Using Monocular Machine Vision

“…These reported results are presented in Table 4. By comparison, Guo and Fu [1], used a simple geometric approach for planar measurements, and tested a 25 mm coin and 35 mm and 85 mm block, which resulted in an accuracy of 0.028 mm, 0.075 mm, and 0.196 mm, and standard deviations in sequential readings at 0.0011 mm, 0.0043 mm, and 0.0083 mm, respectively. Moreover, Li [16] employed a monocular vision system to measure the height and radius of the cylindrical parts of shafts.…”

“…Mechanical parts manufacturers have increasingly stringent requirements for tolerances in their specifications. Meeting these demands means more extensive testing of parts to optimize machine setup parameters, and to ensure quality control [1]. Meticulous manual measurement techniques demand the operator's attention and precision.…”