Gears, as an important component, have always played a huge role in human production and life. Effective detection of various parameters of gears using appropriate methods is of great significance for improving production efficiency and quality.
At present, the parameter detection of gears is slowly transitioning towards automation, shifting from traditional manual detection to non-contact detection. The traditional contact method for measuring helical gears is not only difficult to operate, but also has a high cost and is prone to damage. Non contact detection mainly includes phase shift method and machine vision method. The use of machine vision for parameter detection of gears has the characteristics of high efficiency, high precision, and easy automation, which also makes it play a huge role in gear detection technology.
Xie Duo proposed a basic parameter detection method for spur cylindrical gears, which is based on the Canny operator to extract the sub pixel contour of the gears and detect their parameters. Guo Zhongfeng studied a geometric parameter detection system for spur gears based on Matlab, fitting the tooth top circle and tooth root circle to determine their basic parameters. Zhi Shan conducted research on the detection method of spur gears with small and medium-sized modules, and conducted high-precision detection of local visual images of gears. Shi Wei [6] designed a parameter detection method for helical gears based on the upper and lower end faces of helical gears, which determines the helical angle of helical gears by obtaining gear end face images twice.
At present, most of the research on gear parameters focuses on the basic parameters of spur gears. Therefore, in response to the difficulties in parameter detection of helical gears, such as noise in end face detection and helical angle detection of helical gears, reasonable selection of light sources is used to improve image contrast and reduce noise. After preprocessing, threshold segmentation, and edge detection, accurate helical gear end face contours are obtained, and the basic parameters of helical gears are detected. The support vector machine was applied to the rotation detection of helical gears, achieving classification detection of left and right rotation of helical gears. A spiral angle detection method is proposed, which calculates the inclination angle of the tilted tooth top surface relative to the horizontal axis after obtaining the tooth top area of the helical gear. The spiral angle is converted by the relationship between the tooth top circle spiral angle and the indexing circle spiral angle. The feasibility of this method has been verified through experiments.
A detection method for the geometric parameters, rotation direction, and helix angle of helical gears has been proposed for parameter detection. A machine vision measurement system was built, and the tooth top circle, tooth root circle, indexing circle, tooth number, and modulus of the helical gear were detected and calculated through image preprocessing, threshold segmentation, and edge detection. SVM algorithm was used to detect the rotation direction of the helical gear on the gear side, utilizing the geometric relationship between the tooth top circle helix angle and the indexing circle helix angle, Measure the inclination angle of the tilted tooth top surface relative to the horizontal axis and convert it to obtain the spiral angle of the indexing circle. In the experiment, there were no errors in the detection of the number and modulus of helical gears. The average relative error of the tooth top diameter was 0.59%, the average error of the tooth root circle diameter detection was 0.27%, the accuracy of the rotation detection was 98%, and the average error of the spiral angle detection was 0.49%. It has good accuracy and reliability, has certain theoretical and practical value, reduces the cost and error of manual detection, and can be further applied in online detection.