Study Of Visual Identification System Of Pointer-Type Precise Pressure Gauge

Pointer-type precise gauge is widely used in the petrochemical engineering,electric power industry and other fields because of its simple structure and high preciseness with extra advantages of waterproof,dustproof and anti-electromagnetic interference etc.At present,the calibration of pointer-type precise gauge mainly depends on human vision.In order to improve the efficiency and accuracy of the calibration work,this paper has designed a set of calibration system to auto-recognize the pressure reading based on the present pointer-type precise gauge in the petrochemical industry.Firstly,a set of image acquisition system is built on machine vision with the method of strict hardware selection.CCD camera with resolution of 1600x1200 instead of human visual system is selected to collect high-quality images.Bowl shaped light source is chosen to guarantee that light can be acquired sufficiently and avoid that ambient light possibly disturb the acquisition system.Furthermore,industrial camera and light source can be moved to proper position via displacement platform to guarantee the adaption of the image acquisition system to pressure gauge of different sizes.Secondly,there are some flaws for present vision calibration algorithm.This paper mainly focused on the picture processing algorithm during calibration process.Picture enhancement processing with selected pictures is aimed to weaken image noise and strengthen picture contrast,so that to reduce the subsequent calculated amount.Image edge extraction algorithm and circle fitting method are used to position the point center and also obtain the fitting circle of the dial plate.Followed is the scale line reading based on the central projection method.Then error correction and further completion of missing scale data can guarantee the preciseness of the whole scale reading extracted.Next is to obtain the fitting lines with the method of Hough transformation.All in all,this paper has proposed an improved distance method to acquire point reading combined with present angle and distance method.Lastly,as a sold base of all methods above,the author has succeeded in the preprogramming and operation of machine vision software under the circumstance of Windows.But in considering of compatibility,the author has programmed a set of picture processing software with the help of C++ based on Visual Studio 2012.In comparison of human vision,this algorithm this paper proposed has advantages of high prescission and good stability so that it can meet the requirements of the calibration processing of pointer-type precise gauge.