Research On Geometric Feature Information Locating Algorithm And Its Application In Gear Measurement

With the development of intelligent technology and computer technology the application of machine vision technology is more and more widely popularized.As an example,we can achieve the goals of quality inspection,character recognition,positioning and other functions.Gear,as a paramount transmission part,is required exceedingly high in precision but,now its accurate parameters can be obtained by using machine vision and image processing.This paper focuses on the key techniques of visual positioning of gear measurement,image acquisition by using image acquisition card,processing gear source image by filtering denoising,threshold segmentation and shape description and applying various algorithms to the development of image processing software for the purpose of processing gear source image.The edge detection of object can be realized by using the developed geometric feature algorithm and the detection range and becomes more expanded by dynamic measurement location,in addition the circle in the PCB board is used to verify the algorithm.Finally,all gear parameters can be realized accurately,and we have created an automatic system of measuring gear parameters.The main works of this paper are as follows:(1)Analyzing the importance of gears in industrial production and the current situation of measurement technology at home and abroad.According to the limitation of actual conditions,system requirements and selection rules,the selection of camera,lens and light source of the system is completed.What is more,calibrate the camera.The overall design plan of the system software is completed.(2)The paper analyses and studies various image processing techniques for gear images applying the filtering algorithm,threshold segmentation,edge detection and shape description compared with different processing methods of the gear images.The optimal algorithm is selected and the median filter and gaussian filter combined with filtering algorithm and maximum entropysegmentation algorithm for gear preprocessing algorithm in order to further discard the noise in the image,based on regional characteristics of the screening,and the design of the image processing software is completed.(3)AS to the precise positioning of the gear's top circle and tooth root circle,the traditional Hough transform can not reach the required precision.we have realized the edge detection based on the vertical feature of the object edge,obtained a series of edge data points and adopted an improved random sampling algorithm for edge fitting.The template matching is used to realize dynamic measurement for the initial positioning in order to increase the edge detection range.Using the circular area on the PCB to verify the algorithm,the experiment shows that the detection algorithm can measure the edge of the object well.(4)The goal of measuring a series of parameters of gear has been achieved.The methods above,the top circle and the tooth root circle are contained.Then the tooth number image is obtained and the number of teeth is measured by the connected domain count.Other basic parameters are also obtained by using a series of related formulas.Secondly,the gear pitch deviation,tooth profile deviation and common normal are obtained.Later,a multitude of gears were used to verify the algorithm and the results shows that the algorithm can measure precisely the parameters and the error is within a certain range.(5)In order to realize the acquisition of gears,the calibration of camera and the measurement of gear parameters,a set of software of gear measurement are developed.