Research And Application On Recognition Methods Of Automobile Dashboard Readings Based On Machine Vision

Automobile Dashboard is an important part of the car, as the car’s information displaycenter, automobile dashboard is an important tool to help the driver get vehicle running status.Along with the rapid development of automotive electronic technology, more and moreinformation appears in the automobile dashboard. The logic of automobile dashboard isbecoming more and more complicated. Increasing function makes automobile dashboardfunctional testing even harder. Traditional testing methods can not meet the test requirements,the use of machine vision technology makes it possible to automate functional testing.Using machine vision technology to read the automobile dashboard intelligently brings alot of benefits. This methods can avoid the error caused by fatigue or observationdistance transform, which improves the efficiency and accuracy of identification. Meanwhile,the intelligent recognition system liberates testers from the heavy work of repeat tests, thetesters can focus their attention to write test cases, improve the chances of finding bugs, savingthe labor cost of automobile dashboard test.The main purpose of this paper is to explore the automobile dashboard intelligent readingmethods, which depend on machine vision algorithms. Create an automated test system to findout the effectiveness and availability of the recognition algorithm. In this paper, you can see thefollowing works.1. Image processing algorithms and Hough transform is used to detect the line of meterpointer, and then use the angle conversion method get the value of auto meter. This recognitionmethod is proved high efficiency and strong anti-interference.2. Using Hu invariant moments as a picture feature to recognize automobile dashboardwarning lights and TFT screen static icon. Using color recognition algorithms and flickerfrequency calculation method to recognize the color of the warning lights, TFT icon and flickerfrequency.3. Using Sift algorithm as the picture feature to recognize the floating icon in TFT screen, this method gets Good recognition effect.4. Using OCR technology to detect the word in TFT screen, use neural network trainingcharacteristics of Chinese characters library which can improve the accuracy of characterrecognition.5. Create an image template database, and write a corresponding database managementsystem.6.Design recognition system’s hardware platform, and completed the preparation of thesystem software relying on the.NET development platform and Emgu CV image processinglibrary.The actual use results prove that the automobile dashboard reading recognition systembased on the above identification methods meets the actual needs, and achieved good effect. Itis more accurate and more efficient than manual calibration tester. Because ofsome reasons, shortages exist in the program. Those shortages of the system are summarizedat the end of the paper, meanwhile, the thesis offers suggestions and prospect for the futureresearch work.