Automatic Measurement Technology For Shaft Parts Based On The Analysis Of Sub-pixel

Traditional measurement methods are Artificial for large batch of spare production. In this way,not only measurement efficiency is very low,but also easy to make some mistake,in this thesis,we can implement the automatic measurement for the surface defects and dimensions of parts based on the machine vision technology.There are two important points for the detection of shaft parts,dimension and surface defects. The key of dimensional inspection is to improve the measurements,su rface defects include pits,scraches and burr,and these defects are very common to inspect on the surface of shaft parts,and the key of surface defects is to extract defect rapidly an accurately.in order to solve these problems,l mainly do the following work.First of all,optimize the structure of the test device.In order to reduce the influence of foreign object on CCD image capture system,Set up lens and CCD horizontal,and put in a LGP between light and shaft parts to make the light evenly,the structure of CCD image capture part is a closed darkroom,and this can avoide natural light disturbance from the outside world.optimize the transport velocity of shaft parts,and this can grab the all surface condition from CCD.Secondly,as for the aspect of image algorithm,I made some improvement.In order to smooth the image,this article apply three different image filter algorithm to smooth the image that has been polluted seriously by the noise,and the three different filter algorithms are the average filter algorithm, the switching median filter algorithm and the adaptive median filter algorithm,and choose the filter algorithm that has the best filter effects as the filter algorithm of this article.Through Compared with the different traditional edge detection algorithm,and use the edge detection algorithm that has the best segmentation result,and improvement Sobel edge detection algorithm with two points,the one is use eight direction template,and the other is refine the edge on the gradient direction,so as to get the single-pixel edge.This article examined three different sub-pixel edge segmentation algorithm,compared their Advantages and disadvantages,in order to get high accurate boundary,this paper improvement the gray moment sub-pixel edge detection algorithm,and this improved algorithm has a high location accuracy and repeatability positioning accuracy.Locate the detection area with the grey value projection,and use different detection algorithm to find out the pits,the scraches and the burr.In the end,use the mentioned detection algorithm to segment the image on the test device for several times,the test object is the bar that has a diameter of 5mm±0.05,and analyse experimental results,it turns out that the mentioned detection algorithm has a good location accuracy,and it can also detect all these defect.