Abstract: Aiming at the problem of how to locate the coordinates of the center point of a circular cylinder part and the abnormal fluctuation of the location data, based on the study of the coordinate distribution of the detection point of the edge point of a circular cylinder part, a method combining the processing of the data outliers by the quartile spreading method and the fitting of the circular curve based on the least square method is proposed. First, the collected image was preprocessed with filtering and enhancement, and then the gray image obtained by the original image conversion was processed with the edge detection algorithm based on Canny to extract the coordinates of contour edge points. To simulate the edge point coordinates in the end of link, the distance of the center of the edge points to the average square to make order of sequence, use first quartile distribution method to screen out edge point coordinates of outliers, reoccupy represented by a sequence of the number of edge points coordinates instead of outliers, the coordinates of the final after using least square method to deal with the coordinates of the data fitting get an accurate coordinates of the center of the circle. The experiment of center positioning system and feeding equipment based on OpenCV shows that this method can improve the accuracy of center coordinate points and the accuracy of feeding equipment.