The Research On Calibration Method For The Ice Sensor Based On Image Processing

The airplane icing phenomenon causes serious imperilment during the aviation. Nowadays, more and more attentions have been paid to research of the aircraft icing detection technique increasingly in the world. At the domestic level, the research and development on the ice detection technique has made significant progress. However, there is still an evident gap between the research and development on the ice sensor calibration method and on the ice detection technique. Therefore, the research and development on the ice sensor calibration method, has a great significance to the comprehensive and further development of the ice sensor, and ensures it's feasibility and reliability. In this thesis, it puts forward a frame of reference, which is the research on the calibration of ice sensor based on image processing. It mainly researches a calibration technique through image measurement technique, and is applied to the calibration for ice sensor developed in our lab. This calibration is to be continuous, real-time and automatic, as to improve the efficiency and reliability.The calibration method, which is based on the image processing and for the ice sensor, processes the sectional image of the ice through modern digital image processing technique, then calculates the thickness of the ice on the ice sensor. The process mainly processing procedures includes following processes. First, smoothes and filters the ice image. The second, detects and extracts the upper and lower edges of the ice layer and the gauges of the staff gauge that qua a standard thickness in the measurement separately by Canny operator. The third, connects the ruptures in the edges and identifies all the edges in the image. Then it calculates the average distance in pixels between the two edges of the ice layer and the distance in pixel between the gauges separately. After transformation according to the position relationship between the ice and the staff gauge, gains the actual thickness in mm of the ice layer, through comparing it to the actual scale on the staff gauge, and finally achieves the calibration for the ice sensor. By the simulations in MATLAB during the development, the results demonstrate the feasibility of the scheme. And the actual practice results have good consistency with the simulations in the edge detection, edge extraction and ruptures connection in the edges. In this thesis, the errors of the calculation are analyzed in details, and puts forward and an improvement to the measurement algorithm.