Font Size: a A A

Research And Implementation Of A Pointer Type Instrument Recognition System Based On Color Reduction

Posted on:2017-04-22Degree:MasterType:Thesis
Country:ChinaCandidate:J LiFull Text:PDF
GTID:2348330512457168Subject:Electrical engineering
Abstract/Summary:PDF Full Text Request
Instrument automatic calibration,an important part of modern intelligent manufacturing industry,directly affects the instrument quality and the expenditure cost.Instrument calibration based on machine vision is gradually displacing manual detection,which makes the calibration technology grown towards fully automated and high precision detecting.However,problems are existed in the intelligence calibration,such as head hard identification,lower recognition rate.In this paper,a pointer meter reading recognition system based on color reduction was designed,which achieved the goals of rapidly identifing the header and accurately calculating the reading while reduced cost.The main contents and conclusions of this paper are as follows:(1)Header identification based on the color reduction methodIn order to solve the problems of difficult to recognize the header caused by small scale losing way to statistic and light affects to color distribution,this text designed a head identifying method based on binary color chart.According to the characteristics of the industrial head with single background and main color not changed under different illumination.The plane of each primary color object is quickly formed by using a color reduction method.By extracting the combination characteristics of scale and pointer,using the method of model matching to identify the header.The recognition rate of this method is about 96%,the accuracy rate is relatively improved comparing with the former method.(2)Accurate recognition of readings based on scale repositioning and pointer local fittingThis thesis proposed a method of scale re-calibration and local positioning pointer to accurately calculate the readings,which aimed to solve the problem of missing graduation,non-numerical value of scale points and deviation of the edge of the pointer from the center line.The method of obtaining the scale spacing based on half-image slope statistics has improved the adaptability of the determination of the division angle.By reconstructing the coordinate system,the method of circularcoordinate calculation based on the one way maximum value has ensured the original scale point / value.The method of straight line fitting based on the centroid local has improved the accuracy of the pointer positioning.The error of the numerical calculation is further reduced by using the method of reference point combined with the readout of the deflection angle.Compared with the previous reading recognition,the error of this method is within ? 0.69%,which improves the accuracy of indication calculation.Each module system was tested on the MATLAB development platform.The results show that the recognition rate of the system is over97%,while a kind of pointer meter with uniform scale,single background,center of rotation coincides with the dial center,and the time is about 3.2s.Also,a testing system based on the C++ language is developed.The research ideas of ??this paper has a certain reference value for the further development of meter verification system.
Keywords/Search Tags:Intelligent Manufacturing, Machine Vision, Color Reduction, Instrument Calibration
PDF Full Text Request
Related items