Study Of Strain Gauge Localization Based On Machine Vision

Strain gauge,also known as resistance strain sensor,has the advantages of simple structure,stable performance and high sensitivity,and is suitable for dynamic measurement.The precision of strain gauge is very high.In order to meet the precision requirements of strain gauge in engineering practice,it needs to be polished in the manufacturing process of strain gauge.This process can adjust the precision of strain gauge and make it meet the requirements.However,the strain gauge location needs to be known in advance for grinding,so this paper studied the strain gauge positioning.This paper takes strain gauge sample as the research object.After analyzing the positioning accuracy requirements of strain gauge,the machine vision algorithm is adopted to locate the strain gauge.First,the strain gauge image is preprocessed and threshold segmentation is used to set the threshold value to make the strain gauge characteristics explicit.The feature noise of strain gauge is filtered by median filter.The connected domain extraction of the preprocessed strain gauge image is performed.In the initial method of this paper,the corner marking part of strain gauge is extracted by means of screenshot,which includes the grouping of strain gauge blocks and the four corner marking of strain gauge around them,and finally the position of strain gauge is located indirectly through the corner marking.Then,relevant morphological theories are added to locate the strain gauge in this paper,and morphological filtering method is applied to remove the image noise of strain gauge.First,Otsu was used to perform threshold binarization for strain gauge images,and then morphological edge extraction was performed for the binarization images.Redundant information was removed and corner points were extracted according to the edge contour,so as to find the required position of strain gauge.Such image processing algorithm has low complexity.Finally,based on the characteristics of strain gauge position information,the minimum external rectangle technique is introduced.The specific steps of the final algorithm are as follows.The algorithmadopted in this paper firstly carries out image binarization,extracts strain gauge groups by using the connected domain extraction method,roughly locates the strain gauge,and then repairs the image using morphology.The image of the part connected with the strain gauge is dispersed and the minimum external rotation rectangle of each strain gauge in the strain gauge is drawn to locate the center of the strain gauge,so as to realize the final positioning of the strain gauge.Finally positioning error in 0.0816 ?,below 0.1 ? error that could satisfy the requirement of engineering practice.