Rsearch On Detection Technolgy Of Automobile Meter Reading Based On Machine Vision

Dashboard reading of the automatic identification technology is an important partin the field of automobile testing. In the car dashboard increasingly beautify and concise,high-precision, multi-function is also the most important requirements for a driver.Currently, machine vision technology gradually is applied to the detection of cardashboards, to avoid the artificial visual judgment of subjective factors to improve theaccuracy and efficiency, but for the complex background information on car dashboards,development of detection technology is not mature enough, there are still false positivesand low precision.Based on the study of machine vision technology, complex for backgroundinformation on the car dashboards, improved the handing recognition algorithm, furthercompletion of the automatic detection of the instrument panel reading automaticdetection, the main tasks are as follows:The study on machine vision technology and application of image processingtechnology in the dashboard reading recognition. Design the overall program ofdashboard readings detection from video capture to reading recognition. Based on theaccuracy requirements of reading detection, select the instrument driver and videocapture device.The testing overall program of the instrument panel readings, processed videoimage will be converted into a continuous set of static image frames, analysis ofredundant information, and extract key image frame. Research the backgroundinformation of each frame, by the color space conversion, get containing a single colorimage layer, complete the image filtering and the image binarization.Based on the mathematical morphology algorithm, marking the connected regionof the binary image, according to the area feature get the pointer and scale lines.By theHough transform algorithm achieve the positioning of the dashboard pointer, comparingthe analysis of different curve fitting methods, chose the cubic spline interpolation to fitthe calibration curve, and complete calibration of the main scale node.Based on matlab software development platform, complete image preprocessing ofinstrument panel, the extraction of the effective region and the calibration of dial pointerand the scale line, combined with the distance method to achieve the final testing ofautomobile dashboard reading, and the qualitative analysis of encounter errors of thetesting process.The experimental results show that the improved auto meter reading recognitionalgorithm based on machine vision for background information on complex dashboard, the readings are accurate and reliable. It basically meets the requirements of practicalwork and application, with some prospects for the development and promotion ofpractical value.