Research On Interferogram Tooth Surface Region Segmentation Method Based On Local Density And Frequency Domain Information

Gear is a basic component widely used in the manufacturing industry.The shape error of its tooth surface is an important evaluation index of gear performance,which has an important influence on transmission performance and service life.The realization of precise measurement of gear tooth surface is to improve its performance.The prerequisite for manufacturing accuracy and repair accuracy.Laser interferometry is one of the development directions of non-contact measurement because of its high theoretical accuracy.During measurement,the interference fringe image is the only data collected so far,and its processing method and processing accuracy directly determine the accuracy and reliability of laser interferometry.This article will conduct an in-depth study on the problem of tooth surface domain segmentation of interference images.This paper compares and analyzes the existing classic tooth surface segmentation algorithms.In this study,commonly used interferogram foreground region segmentation algorithms(including relative modulation method,tooth surface object image gray method and the same group of interferogram joint filter recognition method)are used to extract and segment interference fringe patterns.By comparing the segmentation results,the advantages and disadvantages of different methods are discussed;in the process of using these methods to segment the tooth surface area of the interference image,a threshold selection method suitable for this research is determined;according to the advantages and disadvantages of different algorithms,the application situation is explained and Suggestions for the selection of different interferogram tooth surface segmentation methods are given.This paper proposes an image segmentation correction algorithm based on local density analysis.Since the focal plane is perpendicular to the tooth width direction in the measurement,the incident side and the emitting side of the tooth surface cannot be focused at the same time,and the edge of the defocus side is distributed in a zigzag pattern,which affects the segmentation accuracy of the interferogram tooth surface area.The image segmentation correction algorithm based on local density proposed in this study first generates a two-mask image to assist the segmentation of the tooth surface area;secondly,calculates the local density of the pixels of the binary mask image,and divides the pixels into High-density area and lowdensity area;Then,the number of high-density pixels in the kernel matrix is counted for lowdensity pixels,and the boundary points of the tooth surface area are distinguished according to the statistical results;finally,the selection principle of calculation parameters is given and verified by experiments The feasibility and correctness of the method.This paper introduces the image frequency domain information for the segmentation of interference fringe image tooth surface domain.For a series of stray fringes generated at the edge of the interference fringe image and the noise in the image,the current research mainly focuses on processing the time domain information,and the frequency domain information of the interference image has not been fully utilized.This paper firstly performs Butterworth high-pass filtering according to the spectral characteristics of the interference fringe image to retain the details of the tooth surface edge;secondly,the filtering result is binarized for subsequent boundary extraction;secondly,the interference image tooth surface domain fringes are obtained by wavelet transform After the detailed information is binarized;then,the striped binary mask is interpolated and combined with the binary image of the Butterworth filter to identify the boundary points of the tooth surface domain;finally,the tooth surface is segmented based on the frequency domain information through experiments.Effectiveness and accuracy of domain methods.