Calibration Equipment Of Volumetric Flowmeter Based On Image Processing

In order to ensure the accuracy and consistency of volumetric flowmeter's multiple output signals,an automatic calibration system for volumetric flowmeter based on image processing is studied in this paper,which validates the accuracy and consistency between the pulse count value of transmitter and the readings of mechanical head of flowmeter.The main contents of this paper are as follows:(1)This paper accomplishes the overall system design,which collects images of mechanical head and pulse signals of transmitter in real time and transfers these information to computer.Moreover,the system can recognize the flow reading of flowmeter's mechanical head by image processing technology,consequently,the accuracy and consistency between the pulse count value of transmitter and the readings of mechanical head of flowmeter are tested.(2)On the basis of parameters and working environment of volumetric flowmeter,the acquisition hardware devices of the calibration system for image and pulse signals are selected,which can collect images of volumetric flowmeter's mechanical head and pulse signals of transmitter in real time simultaneously.(3)The collected images of flowmeter's mechanical head are processed,and achieve the recognition of flow indicator in images.First of all,to eliminate the interference information,image preprocessing algorithms,such as image graying,image enhancement and image binarization,are carried out.In addition,background compensation image enhancement algorithm bade on vertical projection is proposed,which increases contrast of image and reduces the effect of illumination on character region.The character region localization is based on AdaBoost algorithm,which can overcome the interferences of pointer,illumination,environment,and improve the accuracy of character region localization for flowmeter's head.The pointer reading recognition algorithm is based on image thinning and Hough transformation,and this algorithm can get higher recognition accuracy for pointer reading.Template matching,KNN and SVM character recognition algorithms are compared based on the extraction of principal component features and statistical structure features of character pixels.The character recognition algorithm based on SVM is adopted to recognize single character image of flowmeter's head,and the recognition rate is up to 99.33%.As to the case that pointer-blocking character area influences character recognition,a character recognition algorithm based on prior knowledge is proposed to improve the accuracy of character recognition.(4)The system software for calibrator of flowmeter's head,which achieves signal collection and processing,is designed to validate the accuracy and consistency between the pulse count value of transmitter and the readings of mechanical head of flowmeter.The system has great application value by lots of experiments with 5 volumetric flowmeters on the spot.