Measuring Method of Involute Profile Error Based on Machine Vision

To meet the requirement of fast on-line measurement of tooth profile error of medium and small modulus spur gear in automotive industry, a new algorithm for measuring involute tooth profile error based on machine vision with gear local image is proposed. The gray image of gear is acquired by gear visual measuring instrument, and the pixel information of transition zone of gear profile edge is extracted by Gauss filtering and threshold segmentation. The center of gear positioning is determined by fitting the circle based on fixed radius least square method with constraints. By extracting sub-pixel edge feature points of gear profile, a visual measurement model is constructed, and the measurement parameters, total deviation of involute profile and shape deviation of gear profile are calculated. The total tooth profile deviation of the tested gear measured by the method in this paper has reached 5-level accuracy. Compared with the result measured by M&M3525 NC Gear Measuring Center, the maximum measurement error of the total tooth profile deviation measured by the visual measuring instrument is $1\mu \mathrm{m}$, the maximum error of average measurement is less than $1\mu \mathrm{m}$. Experiments show that the measurement accuracy of the visual measurement system is equal to that of the gear measuring center, and it can meet the requirements of measuring the total tooth profile deviation of 5-level precision gear.