Research On Pointer Meter Reading Technology Based On Machine Vision

The machine vision technology through the camera to simulate human eyes,get the color,shape,size and other information of the object in the real world and form the image,and then through the computer simulation of the human brain,processing,analysis and understanding of the image,finally to determine the categories and attributes of objects and state.Machine vision technology has greatly promoted the automation and intelligentization of industry and improved the efficiency of production.In the production and practice of society,the use of pointer instrument is very wide.In industry,instrumentation can detect current,voltage,flow,pressure and other physical quantities.On the one hand,it monitors production status,and on the other hand controls production activities in a favorable direction through feedback control.With the rapid development of science and technology,the categories of meters are increasing.The functions of all kinds of instruments have also become very perfect,playing a more and more important role in social production and practice.The production level and quality of instrument reflect the scientific and technological strength of a country and the level of industrial manufacture.At present,the reading of pointer meters is mainly done by human beings,which inevitably brings some problems.For example,repetitive work is not efficient,and the result of reading is prone to bias.On the one hand,reading the number of instruments by manual will undoubtedly increase the production cost of enterprises.On the other hand,these problems will also reduce the production efficiency of enterprises.Therefore,it is of great significance to use machine vision technology to recognize the reading of pointer meters.In this thesis,the automatic reading technique of pointer meter reading based on machine vision is studied.According to the pointer and instrument image captured by the inspection robot camera,the recognition and location methods of instrument dial,the segmentation and recognition method of instrument pointer,the calibration method of instrument calibration line,the corner detection method and the calculation method of pointer reading are mainly studied.In fact,due to the limitation of the environment and shooting conditions,the overall image contrast is small,and the detail information such as the target edge is not clear.Therefore,in this chapter,from the perspective of machine vision,based on the principle of digital image technology,the linear gray level mapping method is applied to contrast enhancement of pointer meter image,so that the recognition and recognition of instrument dial is more accurate.Then the pointer is difficult to accurate segmentation problem,this paper uses piecewise segmentation and recognition method in recognition of two parts center and the center of gravity and tip were extracted from the center and tip center circle as feature points to calculate the pointer angle.Then,to divide the scale circle of the instrument,the method of combining the region segmentation with the edge segmentation is adopted.This method can solve the problem that any one of these methods cannot be used to divide the complete scale.Then,the parallel thinning algorithm is used to refine the scale circle,and the branch point extraction algorithm is used to extract the intersection point between the scale line and the scale circle,to indirectly express the scale line.Finally,the pointer readings are calculated according to the ratio of the angle of the pointer to the total angle of the range,and the algorithm is designed and simulated by MATLAB.