Research On Data Recognition Of Weighing Instruments Based On Machine Vision

Machine vision technology has been widely used in industrial production,scientific research,life services and other fields.In order to determine whether the instrument produced by the company satisfies the industry standards,it is necessary to test the instrument in the Metrology Institute.The digital recognition results of the weighing instruments tested are compared with the standard readings of the industry according to the technical requirements,so as to determine whether the digital readings of the weighing instruments are accurate or not.The process of digital acquisition is usually carried out under the condition with electro-magnetic interference,extreme humidity and temperature,which are usually harmful to human body to some extent when people are close to the measurement device.In addition,people are prone to fatigue failure when observing batch data continuously,which will affect the judgement on the accuracy of the instrument data.Therefore,the development of a system of data recognition of weighing instruments based on machine vision has much significance on research and application.This dissertation focuses on the research of recognition method of weighing instrument measurement data,and develops a set of weighing instrument measurement data recognition system.The feasibility of image processing and recognition algorithm of weighing instrument is verified.The main work includes:The software flow of data recognition is designed.The operation of digital image acquisition,ROI selection,image preprocessing,tilt digital vertical correction,digital segmentation and normalization are realized respectively.Image preprocessing includes gray level,binarization,color reversal,filtering and morphological processing operations.The average tilt angle of all digits in a picture is determined by searching for the smallest outer rectangle of digits.The vertical correction of tilt digits is accomplished by reversing the average angle of all digits in the picture.The normalized characters are recognized.Character recognition based on threading method and BP neural network method is designed and implemented respectively.The threading method divides the number into upper,middle and lower parts with two straight lines passing through the number.The third straight line passes through the number and divides the number equally into left and right parts.The number of intersections between a number and three straight lines is used to determine the value of the number.Although the threading method has a fast recognition speed,it can only recognize seven segments of LED digital tube numbers.The BP neural network algorithm first designs the network,and it trains the designed network with batch samples until the neural network enters the mature stage of application.BP neural network algorithm can recognize LED seven-segment digital tube number and LCD display number,and has good applicability.The decimal point is generally located in the lower right corner of the number.In this dissertation,the decimal point is judged and recognized by the position characteristics of the decimal point in the number.A set of weighing instrument measurement data recognition system based on machine vision is designed and developed.The recognition effect and the feasibility of the algorithm are verified by the digital pictures of three weighing instruments.The experimental results show that the system can realize the digital recognition of LED seven-segment digital tube and LCD display screen,which shows that the designed machine vision weighing instrument measurement data recognition system is effective and feasible.