Infrared Multispectral Solution To Ice Detection On Flight

For the aircraft due to ice, snow, frost and other pollutants will cause great aircraft flight safety hazard, so that the residual de-icing of aircraft ICAO have established a strict standard, in the winter icing conditions, aircraft de-icing must be carried out before and after take-off detection of residual ice component. This paper take the method that infrared multispectral aircraft ice detection, which uses of the ice and snow of infrared absorption/reflection characteristics, contrasts infrared images under different wavelengths of infrared channels for detection. In order to ensure accuracy, we must request the images under different channel pixels eleven matches. In order to eliminate the mismatch impact which cause by the environment or misoperation, on the basis of today's match algorithm the paper explores and studies the robust matching algorithm in depth. But four images matching has a higher demand in the matching accuracy and matching speed, so we can not directly apply the algorithm with two images matching on the four images, only to improve it.This paper presents an improved RANSAC robust matching algorithm, which adds pre-inspection procedures based on the disparity gradient constraint and the effective area constraints. We can effective reduce the count of calculation error homography, thus eliminating the corresponding homography test session. On the basis of the improved RANSAC matching algorithm, using the correlation between the four images feature points, this paper presents a matching framework aimed at the four images matching algorithm. And increase the matching efficiency to a higher degree, shorten the matching time. Finally, we do the simulation experiment by the on-site image data, and the results of data analysis show that the algorithm is faster than the original one.