Machine Vision Based Design And Implementation Of Automatic Reading System For Pointer Meter

The instrument can be seen everywhere in production and life scenarios,such as electricity meters and water meters.The intuitive,easy-to-record nature of electronic digital meters makes electronic instruments more widely available than pointer meters.However,in some specific situations,such as dust,high humidity,flammable and explosive,pointer instruments are still in large use.The traditional instrument indicates that the number reading depends on the human eye,but the human eye reading will cause fluctuations in the display due to fatigue,so the research on the automatic reading of the meter is widely carried out.Based on extensive reading of relevant research literature at home and abroad,this paper proposes an automatic reading recognition method for multi-pointer instruments with complex background.In addition to making full use of the instrument features,this paper has improved the identification process and identification methods.The main work of this paper is as follows:(1)Design and implement the instrument identification system based on the instrument in the analysis topic.For the subject,it is necessary to select and purchase equipment such as camera lenses,build an instrument identification platform,and repeatedly debug,and try to simulate the real production environment.(2)Perform image processing algorithm recognition reading on the collected sample image.The application of distortion correction,image preprocessing,reading recognition and other methods were carried out on the image.The advantages,disadvantages and applicability of these different algorithms were compared,and the most suitable algorithm for complex instruments was selected.Finally,the subject designed an optimized distance-based reading recognition method,and improved the reading correction method for the pointer occlusion problem of the multi-pointer meter and the missing distance reading scale line problem,and obtained accurate reading results.(3)After accurately identifying the sample image readings,the test plan was designed,and the reliability and stability of the system were evaluated in the test cases.Under the constant temperature condition test,the final meter reading fluctuation range is within 15% of the division value,and the reading error is within 10% of the division value.On this basis,the errors caused by image acquisition and reading recognition are further analyzed,and the possibility of further solving the problem is proposed in the work summary and work prospect.