An Improved Sub-Pixel Edge Detection Method and Application in Measurement of Spring Dimension

Edge detection has been one of the most important techniques in image processing. In this paper, an improved algorithm for sub-pixel edge detection is proposed for the disadvantages of traditional edge detections and sub-pixel edge detections, which is used in the measurement of spring length. Firstly, the mathematical morphology is used to abstract the pixel level edge of spring images, which can keep the intact edge information of spring while restrain the image noise. Then, Zernike orthogonal moments is used to relocate the spring edge to sub-pixel level. At last, the exact spring edge is gained from the discrete sub-pixel points using the least square fitting method. The experiments demonstrate that this method can satisfy both real-time request and high accuracy request.