Grinding wheel contour shape detection method based on machine vision

Aiming at the problem that the grinding wheel dresser can’t detect the contour shape in real time in harsh environment, an inspection system of grinding wheel contour shape based on machine vision is built. The background cut up algorithm based on margin detection is used to set the ROI area, the contour extraction and cubic spline interpolation are carried out in the subsequent images to get a complete contour image, and then the improved sub-pixel edge extraction method is used to get a higher precision contour image. The system can totally compensate the contour curve with illumination variation range from −200 Lux to 150 Lux and contour missing ratio less than 25%, and the time for single image processing is within 160 ms. The average image quality index, square root error of pixel value, structural similarity and image peak signal-to-noise ratio of accuracy detection are 0.550 0, 7 803, 7.24 and 0.037 0, which can realize the detection of grinding wheel contour shape during machining.