基于机器视觉的汽车仪表读数检测技术研究-仪器科学与技术专业论文.docxVIP

哈尔滨工业大学工学硕士学位论文 哈尔滨工业大学工学硕士学位论文 - - II - Abstract Dashboard reading of the automatic identification technology is an important part in 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 car dashboards, to avoid the artificial visual judgment of subjective factors to improve the accuracy and efficiency, but for the complex background information on car dashboards, development of detection technology is not mature enough, there are still false positives and low precision. Based on the study of machine vision technology, complex for background information on the car dashboards, improved the handing recognition algorithm, further completion of the automatic detection of the instrument panel reading automatic detection, the main tasks are as follows: The study on machine vision technology and application of image processing technology in the dashboard reading recognition. Design the overall program of dashboard readings detection from video capture to reading recognition. Based on the accuracy requirements of reading detection, select the instrument driver and video capture device. The testing overall program of the instrument panel readings, processed video image will be converted into a continuous set of static image frames, analysis of redundant information, and extract key image frame. Research the background information of each frame, by the color space conversion, get containing a single color image layer, complete the image filtering and the image binarization. Based on the mathematical morphology algorithm, marking the connected region of the binary image, according to the area feature get the pointer and scale lines.By the Hough transform algorithm achieve the positioning of the dashboard pointer, comparing the analysis of different curve fitting methods, chose the