Scale Image Recognition Of Mercury Thermometer Based On Image Processing

According to the current measurement laws and regulations,mercury thermometers in China must be periodically calibrated to achieve accurate measurement.Thermometer verification is divided into many parts,but the most important part is the automatic recognition of scale image.However,due to the large number of thermometer calibration,the huge number of reading,and the shortcomings of manual reading,such as slow calibration,easy scalding,mercury leakage and so on,at present,a difficult problem facing the work of measurement and verification is how to realize the automatic reading of mercury thermometer.In order to reduce the work intensity of metrological inspectors and improve the accuracy of reading,the current situation of verification needs to be improved urgently.According to the characteristics of mercury thermometer and the working environment of actual laboratory inspection.Research on Mercury Thermometer Scale Image Recognition Based on Image Processing Technology.Firstly,the module,installation and configuration of OpenCV are introduced,and then the pretreatment technology in the process of mercury thermometer scale recognition is deeply analyzed.Including image graying,filtering,threshold segmentation,etc.In order to get better effect and reduce halo,bilateral filtering is used to estimate illumination image.Gamma algorithm based on the change of local area information of image is used for correction,and Retinex enhancement is adopted,and multiple thresholds are matched.According to the characteristics of scale line in thermometer image,image processing of thermometer is carried out,the image of the scale line is extracted and refined to get the position of the scale line,and the scale is projected in two directions to optimize the uniform distribution of the scale and further optimize the purity of the scale.According to the characteristics of the liquid column in the thermometer image,the position of the liquid column head is extracted.The last step,connected domain filtering is used to extract digits.According to the characteristics of reading,the idea of template matching is adopted.When matching digital features,the contour pixels and projections are combined,and the weights of them are allocated according to different fonts,so as to expand the compatibility of digital recognition and make the recognition effect better.Combined with the above extraction part,through calculation,the correct indication of the thermometer is finally obtained.Finally,the application research of thermometer scale image recognition is carried out.The results of many experiments show that the mercury thermometer scale recognition algorithm in this paper has better performance,faster operation speed and higher accuracy.This method improves the verification efficiency of mercury thermometer and has good effect in practical application.