A method and system for detecting and tracking a landing mark of an aircraft

Abstract

The invention relates to the field of image processing, and discloses a method and system for detecting and tracking a landing mark of an aircraft, so as to realize the accurate and fast detection andtracking of the aircraft landing mark and assist in completing the accurate landing of the aircraft. The method of the invention comprises the following steps: acquiring the information of two or more image frames collected by a video detection device on an aircraft, determining a target area corresponding to a landing mark in an initial image frame, then determining a smallest external rectangleof the target area, and determining four corner point coordinates of the smallest external rectangle through a library function; taking the next frame of the initial image frame as the current imageframe, calculating the affine transformation matrix between the initial image frame and the current image frame, predicting the position of the four corner coordinates in the current image frame according to the affine transformation matrix, and marking the predicted rectangular block diagram in the current image frame; taking the current image frame as a new initial image frame, re-determining the current frame, and repeating the above steps to complete the tracking of the landing mark.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a method and system for detecting and tracking aircraft landing marks. Background technique [0002] With the rapid development of the computer industry, the target detection and tracking technology based on computer vision has also been greatly improved. At present, the target detection methods mainly include edge-based corner feature detection methods, circle feature-based detection methods, The detection method based on invariant moments, and the target detection method based on deep learning; the target tracking method mainly includes the method based on template matching, the tracking method based on the region, the dynamic contour tracking algorithm, and the particle filter tracking algorithm; among them, the edge-based Although the corner point detection method has high precision and good stability, the corner point has a great dependence on the edge extraction algorithm, a...

AI Technical Summary

Problems solved by technology

[0002] With the rapid development of the computer industry, the target detection and tracking technology based on computer vision has also been greatly improved. At present, the target detection methods mainly include edge-based corner feature detection methods, circle feature-based detection methods, The detection method based on invariant moments, and the target detection method based on deep learning; the target tracking method mainly includes the method based on template matching, the tracking method based on the region, the dynamic contour tracking algorithm, and the particle filter tracking algorithm; among them, the edge-based Although the corner point detection method has high precision and good stability, the corner point has a great dependence on the edge extraction algorithm, and the broken edge line has a great influence on the corner point detection; It depends on the accuracy of the graphic contour data, and the amount of calculation is large; the target detection method based on the invariant moment can detect the target well, but the computational complexity of the invariant moment is high, and it is only suitable for processing simple two-dimensional image

The tracking method based on model matching can be applied in the complex environment and the situation where the target pose changes, but this method needs to calculate a complex model, and the calculation amount is very large; the area matching algorithm has a large amount of calculation, and in the case of occlusion and light intensity changes, This method may fail to track; the disadvantage of the dynamic contour tracking method is that it is sensitive to the initial value of the target. Once the initial target contour parameters are detected incorrectly, the subsequent tracking process will be completely invalid.

The particle filter algorithm itself is relatively complex, with a large amount of calculation and a large storage space; the target detection method based on deep learning has a very high accuracy rate, but the increasingly high-dimensional feature description brings a burden to the calculation, and the classifier model The pre-training of requires a huge amount of data, and one model only corresponds to one target situation

Images