Research On Sub-pixel Edge Detection Of Small Module Gear Image Based On Moment Methods

The small module gear is the main component of the transmission,and its accuracy directly affects the transmission efficiency and working performance of the equipment.Gear parameter measurement method based on visual inspection technology overcomes the limitations of contact measurement in small cogging clearance and poor gear stiffness,and has the advantages of non-contact and high precision.Sub-pixel edge detection algorithm is an important mean to improve detection accuracy in small modulus gear vision measurement system,which has a great research significance.This paper proposes corresponding solutions to the problems of small module gear image in edge detection.The main research contents are as follows:(1)Researched on image preprocessing technology.This paper studied on the image filtering,morphological processing and pixel-level edge detection algorithms which was needed in image preprocessing.The best image preprocessing scheme was selected for sub-pixel edge detection algorithms through experimental comparison.(2)Researched on Zernike moment sub-pixel edge detection algorithm.In this paper,this paper research on Zernike moment algorithm which was wildly used in the sub-pixel edge detection algorithm.This paper analyzed the influence of the asymmetry in odd template convolution and the limitation in center of convolution window,and proposed an improved Zernike moment sub-pixel edge detection algorithm.Experiments between the proposed and the others have been realized to verify its performance based on simulative and real images.(3)One-dimensional gray-scale moment sub-pixel edge detection algorithm.Focus on the accuracy of Zernike moment sub-pixel edge detection algorithm has limitations,and the gray-scale gradient at the edge of small-module gear image has double-peak or multi-peak distribution,aone-dimensional gray-scale moment sub-pixel edge detection based on neighborhood characteristic analysis was proposed.Experiments between the proposed and the others have been realized to verify its performance based on simulative and real images.(4)Sub-pixel edge detection accuracy evaluation.First of all,the laser interferometer was used to convert the displacement of the platform into the length of the image,and then the system calibration coefficient was obtained by standard gauge calibration.Then,the sub-pixel edge detection accuracy was quantified by the ring gauge image and the small module gear image.By experiments,the accuracy of detection in actual images by different sub-pixel edge detection algorithms and pixel-level edge detection algorithms was analyzed.The research result of this paper is the supplement and perfection of the sub-pixel edge detection algorithm,which contributes to the supplication and development of machine vision in the field of fine-pitch gear measurement.