Studies On Pointer Instrument Recognition Technology

Instrument detection is a common detection method in industry.Common instruments can be mainly divided into electronic digital instruments and pointer instruments.In the detection and recognition process of the entire pointer instrument,the area to be identified is mainly the dial and the pointer of the instrument.The primary premise of the entire image processing is to accurately extract the position of the instrument dial to ensure the subsequent extraction of instrument data.Accuracy,so the dial positioning technology has become an important part of instrumentation testing.In the pointer instrumentation detection and identification of this thesis,in view of the typical geometrical features of the instrument is a circular contour,an automatic positioning technique based on a circular contour is used to realize the positioning of the instrument and instrument position,ie,the Hough circle transformation.The method has the advantages of simple and convenient implementation,small amount of calculation,simple and easy to understand principle,less cost and high precision.This thesis mainly studies the reading test of various pointer instruments,including some special instruments.The oxygen pressure table selected in this thesis,due to the specificity of the pointer structure,directly extracts the first and the last two segments from the Hough extraction.Therefore,we need to remove the interference when extracting.Here,we adopt the length screening method to extract correctly.By extracting the length information of the detected straight line and then retaining the longer needle portion,the shorter tail portion is filtered out.For a multi-pointer instrument such as a clock,we have rewritten the length-related functions of the pointers,and first classify the extracted straight lines to obtain three different types of straight lines.Average the length of each type of line to get three different length values.Then average the deflection angle of each type of straight line to obtain three different deflection angles corresponding to it.The size of the three acquired length values is sorted.The clock studied in this thesis is the modelwith the shortest hour hand length,second minute hand,and longest hand second hand.Finally,the obtained value should also be rounded down.The main method is to perform image preprocessing on the acquired instrument,mainly including graying,histogram equalization,bilateral filtering,and edge detection extraction to obtain the outline of the image.Hough transform the image to find circular outlines and exclude unqualified outlines.Then the pointer of the instrument is positioned by the Hough line transformation,and the useful information such as the dial and the pointer is extracted.After extracting the dial and pointer information,it is necessary to convert the angle between the pointer and the horizontal line into the scale value of the pointer through some angles conversion,and then read out the scale value.This thesis selected some single-point meters for readings,including some gravity meters,pressure meters,and temperature reading meters.And real-time detection of dynamic video pointer readings.A clocks is chosen for readout detection.Finally,the results obtained in this thesis are analyzed and compared with the existing methods.This software has the advantages of high speed,small experimental error,and real-time detection.