Sub-pixel Gear Parameter Measurement Based on Zemike Moment

Gear is a kind of mechanical transmission components used widely in varieties occasion. improving the machining precision of gear has always been the goal of gear manufacturers and researchers. measurement accuracy for gear has important implications for the production of it. Aiming at the disadvantages of high cost, difficult operation and low efficiency of traditional gear measuring equipment, this paper proposes a machine vision gear measuring system running on embedded equipment Raspberry Pi 3B. In order to improve the measurement accuracy and computational efficiency, we use a sub-pixel edge detection method based on Canny edge detection and Zernike moment method, and realize the convolution acceleration calculation process of Zernike template through the GPU integrated in Raspberry Pi 3B. Finally, we performed a gear measurement experiment on this measurement system. By comparison, we found that the error range based on subpixel edge detection is within 30um which can meet the measurement of the gear with accuracy level 7 to 12. Therefore, this system has certain practical value.