The Research On Computer Vision Technology In Automatic Examination System Of Dial Gauge

Computer vision is known to obtain information from image and simulate human's vision taking advantage of computer technology and optoelectronic technology etc. Recently, with the development of computer vision theories, some dial gauge automatic examination systems based on image processing have been brought to use. However, the system of meter calibration in corporation machine, light and electronics has bottleneck of technology. Moreover, meter manufacturer and metrological service press for the advancement of processing speed, higher precision and the betterment of control technology.In order to improve the advancement of processing speed and precision, this paper studied key image processing algorithm in dial gauge automatic examination based on computer vision. On the basis of studying and trying the image processing algorithm, the better algorithm of image segmentation and method of getting parameter in Hough transform were brought forward. Then I built the image processing system of dial gauage pointer recognition and finished the software by VC++6.0. The experiments of angle reading and elapsed time of image processing indicate that the image processing system of dial gauage pointer recognition is able to exactly recognize 1/10 of a scale of dail gauge( 1μm ), detect one pointer in less than 180ms, and realize higher precision and near real-time examination.In this dissertation, the theory, research actuality and development trend of visual inspection are mainly discussed. Aiming at the characteristics of dial gauge inspection, this dissertation discusses the project as a whole and focuses on the reach of the following 3 sections: one is to create rational motion control system for the requirement of automatic examination. The second is to build rational realizing hardware platform in desire of the real time. The third is to study and design the software of the image acquisition and processing algorithm in order to achieve higher precision. The main work in this dissertation is as follows:1. Design and programme software of motion control.2. Discuss methods of choosing image acquisition equipment and design the software of image capturing. Based on apriority of image, effective image processing methods are advanced, which can recognize characters such as pointer and scale, obtain accuracy index and realize the real-time examination.3. The software design of hough transform based on FPGA is finished with Verilog HDL language in QuartusII5.1 and try to get the simulation results.4. As for the image calibration system always have measurement errors and factors of restricting processing speed, this paper also analyzes the reasons for errors, presents the methods of erasing errors and effective ways to increase speed.