Research On Automatic Reading Methods Of Pointer Instrument Based On Machine Vision

Because of its advantages of convenience,high reliability and low price,the pointer instrument is preferred to be used in the field of industrial measurement widely as a kind of measuring instrument.Considering that the industrial field generally needs to perform reading operations on the pointer instrument frequently,the traditional manual reading method has problems such as poor reading accuracy and being easy to be affected by subjective factors.Thus a more efficient way needs to be explored.In recent years,with the rapid development of computer technology,machine vision method has been widely used in the field of industrial measurement.The application of machine vision technology to the automatic reading of the pointer instrument has the advantages of high reading accuracy and high efficiency compared to manual reading.However,in the process of image acquisition for pointer instrument in industrial measurement field,the image of instrument is easily affected by uneven illumination and random change of shooting angle,which leads to the use of machine vision technology for efficient and accurate automatic reading of the pointer instrument still a challenging problem.In order to solve this problem,this paper will focus on the three aspects of instrument panel scale segmentation,disk key parameter extraction,and instrument maximum range identification,as follows:(1)Instrument panel scale segmentation.In order to accurately extract the scale lines of the panel area,a method of instrument panel area localization based on Hough circle detection is proposed.On this basis,the improved Niblack algorithm is used to perform image binary segmentation on the instrument panel area.Then,the scale lines to be identified are accurately extracted based on the connected components analysis method and the geometric features of the instrument scale lines.(2)Extraction of key parameters of disk.In order to realize automatic reading of pointer instrument,it is necessary to obtain some key parameters of the disk,namely the center of the circle,radius and pointer.Based on the extracted scale lines of the disk,the method of RANSAC interior point estimation is proposed to determine the center and radius of the disk.At the same time,considering the longest length of the instrument pointer in the disk area,based on the segmentation result of the instrument panel,an image region accumulative projection method is used to quickly determine the position parameters of the pointer.(3)Maximum range identification of instrument.In order to adapt to the reading of different ranges pointer instrument,it is necessary to automatically determine the maximum range of the instrument.Considering that the maximum range number often appears near the end of the disk scale line,the image region containing the maximum range number can be preliminarily extracted through the end scale line.On this basis,the method of character segmentation,Radon tilt correction and other methods are used to accurately propose the characters to be recognized and the maximum range number can be quickly identified by the template matching method.In this paper,a large number of industrial field pointer instrument pictures are collected to test and verify the proposed method.The experimental results show that the proposed method can successfully solve the problems of uneven illumination and random changes of shooting angle,and the reading accuracy of different ranges pointer instruments is higher than 95%.