Study On Automatic Test System For High Precision Pointer Instrumeng Based On Machine Vision

Pointer instrument is widely used in the field of measuring. The instrumentswill produce certain error after long term of using, in order to ensure the accuracy ofthe measurement, it should be calibrated and accredited in time. At present, thetraditional measuring method is artificially reading the test, while manualverification method is still affected by the viewing angle and viewing distance andthe accuracy of this method is bad. On the other hand, after long-time use of eye, itis so easy for the inspector to produce visual fatigue which has a great influence onthe accuracy of the measurement.In this dissertation, machine vision technology is applied in the system ofpointer instrument verification, using image sensor, computer and other machinesinstead of verification officers’ eyes, this can not only eliminate fatigue of eyescaused by the long-term work, also can improve the efficiency and accuracy ofverification. This dissertation is mainly accomplished the following:The dissertation has completed the overall design of the automatic verificationsystem and made clear the requirements of the overall design. Then, analyzing anddiscussing how to choose the optimal environment, the image acquisition deviceand motion control device. Selecting LED as a light source and using scatterlighting system. Utilizing digital CMOS camera with USB interface as the imageacquisition device, then capturing the images by using specialized softwareLabVIEW. Utilizing KSA electric displacement platform and MC600controller asthe control device.The dissertation has accomplished the image preprocessing operation of theinstrument. Based on the characteristics of the dial’s image, using space correctiontechnology to correct the distortion of the image, using the spatial geometrictransformation method to adjust the distortion of the image, using the methodcombine gray equidistant interpolation with local histogram equalization to achievethe enhancement of image, and enhance the contrast of useful region andbackground region, using function NI LowPass VI to filter the image, using themethod named Otsu to filter the image. The dissertation has completed feature extraction and interpretation ofindicated value from the instrument dial image, and improved the refinementalgorithm used in morphological operations, using the m-connectivity to eliminatethe ambiguity of the image, using clipping algorithm to eliminate the glitch in therefined image and using the improved m-connectivity and improved clippingalgorithm to refine the pointer and the dial scale, making sure the pointer and dialscale can be approximately considered together by one pixel. Using the addoperation, subtraction operation and XOR operation of subtraction method andconnected domain extraction algorithm to extract the features of pointer and scaleline, using Least Squares to fit a straight line to extract the line of pointer and scaleline. In the end, the system can extract the value of the pointer by calculating thedistance between the pointer line and zero scale line.The dissertation has implemented the algorithms and processes in the softwareplatform of LabVIEW, and experimented the reading of the instruments. The resulthas proved that the algorithms in the dissertation are right. And the methods in thisarticle have a certain application value.