Roundness Detection of End Face for Shaft Workpiece based on Canny-Zernike Sub Pixel Edge Detection and Improved Hough Transform

In order to improve roundness detection efficiency and accuracy of the shaft end face, a detection method based on improved Canny-Zernike Sub-Pixel Edge Detection and improved Hough Transform is designed and implemented, which overcomes the shortcoming of low accuracy using single Hough Transform Detection. Firstly, the Optimized Canny Operator is used to extract the pixel level edge gradient of the original image, and then relocate the extracted edge at sub-pixel level along the gradient direction with Zernike Moment. Finally, the Improved Hough Roundness Detection Algorithm is used on the sub-pixel edge, and the roundness error is evaluated with the Minimum Region Method. Compared with the traditional pixel level Roundness Detection Algorithm, it has higher positioning accuracy, stronger noise immunity. The accuracy reaches 0.001 pixel level and detection speed reaches millisecond level.