Gear is one of the most important parts in modern industrial production. In mass production of gears, on-line measurement and monitoring of gear machining accuracy is of great significance to guarantee the processing quality.In order to improve the efficiency and accuracy of gear measurement, measurement technology has gradually changed from traditional contact measurement to non-contact measurement.Machine vision-based gear measuring method not only achieves non-contact measurement, but also has the advantages of low cost, good real-time performance and high automation, which can greatly reduce the labor intensity of the tester and avoid the misjudgment of the measurement results caused by subjective factors.
The general step of measuringby machine vision is to collect image online first, preprocess image and detect edge. Finally, feature analysis is carried out on the obtained edge, and a series of algorithms are designed to calculate gear parameters and errors.The key and foundation of gear parameters and error calculation is edge detection of image, and the classical algorithm of gear edge detection is studied.Traditional image edge detection method is to binarize the image and obtain integer-order edges, such as edge detection algorithm still stays at the pixel level.However, in edge detection of mechanical parts, especially in dimension measurement applications, integer-level edge detection can not meet the requirements of image processing.
Subpixel edge detection algorithm is a high-level edge detection algorithm, which can be used for measuring part size error.In this paper, Zernike moment sub-pixel edge detection algorithm is used for reference, and the online image of spur gear is processed by combining the maximum class-to-class variance method of image histogram, which improves the efficiency of edge detection and reduces errors.Aiming at the disadvantage of coarse pixel width of Zernike moment sub-pixel edge, the collected image is processed with morphological filtering to refine the edge. Based on this, a series of algorithms are designed and developed, the basic parameters of gear and pitch deviation are obtained, and error analysis is carried out.
The basic parameters and machining errors of spur gears measured non-contact with machine vision technology are analyzed and calculated, and the basic parameters and pitch machining errors of spur gears are obtained.
Grayscale and histogram equalization processing are carried out for online acquired spur gear photos. The application of Zernike moment method in sub-pixel edge detection is analyzed. The refined expression algorithm of gear contour is obtained by combining Zernike distance method and morphological filtering method.
The basic parameters and pitch errors of spur gears are reversed by using the designed algorithm.The result of reverse calculation shows that the error of basic gear parameters is0.02%Within this range, the pitch error is also about 5%.It shows that the measuring algorithm designed in this paper is accurate and feasible, and realizes real-time, non-contact and high-efficiency measurement of gears.