Research On Recognition Method Of Aircraft Airspeed Scale Based On Image Processing

In the cockpit of modern aviation aircraft,the way of instrument reading is realized by manual visual reading.When performing manual tests,it consumes a lot of time and consumes too much human resources,and the test accuracy cannot be guaranteed.Computer digitization technology and various electronic network technologies are constantly developing rapidly.In the field of military aviation,video image processing technology is introduced to study the images displayed by the main instruments of the cockpit,so that the images can be automatically identified and detected.Effectively improve the efficiency and accuracy of the test.It is a general trend that manual testing will eventually be replaced by automated testing.Therefore,this article focuses on the automatic recognition and reading system of the airspeed gauge in the main display of the cockpit of a certain type of military aviation early warning aircraft,and the airspeed of the main display in the cockpit is studied.As the research object,the ruler collects aviation instruments in real time through the Gig E Vision protocol,and then uses image processing technology to enhance the collected instrument images by MSR(Multi-Scale Retinex)algorithm,preprocess the enhanced instrument images,and finally use An image recognition method combining Tesseract-OCR and Long Short-Term Memory(LSTM)for automatic character recognition.The main research contents are as follows:(1)In response to insufficient light,rainy or foggy weather,and the presence of obstructions that affect the image quality of aircraft instruments,this paper proposes an MSR-based image enhancement algorithm that adjusts the brightness range of the dark area and introduces The adaptive center surround function.Theoretical analysis and experimental comparison show that the algorithm conforms to the human visual characteristics,effectively enhances the details of the instrument image,solves the problem of color distortion after enhancement,and avoids color distortion.This algorithm lays a good foundation for subsequent identification.At the same time,the algorithm is easy to implement,which is conducive to the integration of the whole machine,and forms a certain type of military aviation early warning aircraft with low power consumption and miniaturization,which has certain economic value and social significance.(2)Through the study of the scroll wheel display problem of the airspeed scale in the primary flight display(PFD,Primary Flight Display),aiming at the scroll wheel digital display characteristics of the airspeed scale in the PFD,incomplete characters appearing in the process of numerical value change are targeted The character set training solves the problem of low recognition rate of Tesseract-OCR when it contains incomplete characters.On this basis,it is optimized,combined with Long Short-Term Memory(LSTM,Long Short-Term Memory)neural network model,experiments It shows that when using Tesseract-OCR to recognize aircraft instruments with incomplete characters,when using the training library,the average recognition rate is 80.61%higher than that of the built-in library.After optimization,the average recognition rate is increased by 3.35% compared with the training library.When there are incomplete characters,the average recognition rate is 79.45% higher than that of the built-in library.After optimization,the average recognition rate is 4.75% higher than that of the training library.This paper proposes a recognition method and implementation of the PFD of the aircraft main display based on Tesseract-OCR.Through experimental tests,the average accuracy of recognizing incomplete characters and only incomplete characters in the aircraft instrument is improved from 5.6% and 1.25% to 89.56% and 85.43%,effectively improve the recognition accuracy,and solve the problem of low recognition rate when Tesseract-OCR recognizes incomplete characters.