| The instruments of Aeronautical Industrial aircraft show the real-time flight status of the aircraft,which is an important basis for judging flight status.With the widespread use of glass cockpit,it has become the mainstream trend of modern aircraft design to replace traditional electronic and mechanical instruments with LCD multifunctional display screen as virtual instrument.In the production test stage,how to quickly and accurately verify the correctness of the virtual instrument screen display data is the key research content of the current aviation virtual instrument data detection technology.In this paper,based on machine vision technology,the key technologies of automatic detection of display data of aviation virtual instrument screen are studied.The main work is as follows:1.Design the detection scheme of the display data of the aviation virtual instrument screen,divide the whole detection process into three aspects: image correction segmentation,pointer instrument detection and digital instrument detection,and design the detection process of each aspect separately.At the same time,study the key technologies involved in each process,and complete the detection with the specific aviation virtual instrument as the target to be tested.2.In image correction and segmentation,PnP algorithm is used to calculate the pose relationship between the camera and the target screen,and then the pose parameters are used to correct the image,which solves the problem of the influence of the pose relationship between the camera and the target screen on image processing in visual detection.The traditional DLT algorithm is improved by RAC model,so that the camera pose parameters can be acquired quickly and accurately.The algorithm adds constraints to DLT algorithm by adding radial constraints,which not only keeps the traditional DLT algorithm low time-consuming,but also improves the accuracy of DLT algorithm for camera pose estimation.Then compared with the commonly used pose estimation algorithms,the accuracy,real-time performance,anti-noise performance and actual correction application of the algorithm are experimentally compared.Finally,it is proved that the algorithm designed in this paper can not only estimate the camera pose quickly and accurately,but also has better adaptability to noise.After image correction,the image is divided according to the pre-designed location area to complete image region segmentation.3.In the aspect of instrument detection,the process of pointer extraction-pointer center line detection-pointer reading is designed for pointer instrument,while the digital display instrument completes the detection by locating character region-dividing single character-character recognition.The related technologies involved in each step are studied and analyzed respectively.Finally,the pointer region is extracted in the form of difference method combined with connected area method according to the characteristics of the sample aeronautical virtual instrument provided by this topic.In the aspect of pointer reading,after extracting image edge with Canny operator,edge straight line is detected by Houghtransform,and the center line of pointer instrument is extracted by least squares fitting straight line method.The center line is used to complete pointer reading and realize the detection of pointer instrument.Bidirectional projection combined with connected region method is used to locate and segment characters,and SVM classifier is used to segment characters.Recognition,the detection of digital display instrument is realized.The experimental results show that the visual measurement algorithm for the display data of the aviation virtual instrument can complete a recognition and detection within 0.2seconds,and it also has a good accuracy when the image is slightly blurred,which has certain practical value and development prospects. |
PDF Full Text Request