UAV Flight Attitude Acquisition Algorithm Based on Multi-Camera

In this paper, an algorithm of UAV flight attitude acquisition based on multi-camera is proposed. Compared with the previous work, this method can realize the recognition and attitude capture of external flying objects and small targets. An algorithm is constructed to capture the flight attitude of UAV by building a camera interface database, and the whole interface pixel database is obtained by regularly calibrating the pitch angle and yaw angle data of a small number of pixels. The whole interface pixel database is obtained by Gaussian dynamic threshold filtering and interpolation. The attitude position of the UAV can be determined by the pitch and yaw line of sight rendezvous of the multi-camera, and the spatial position error caused by the error between the interpolated pitch yaw data and the real pitch yaw data can be expressed by the spatial position offset. In the flight test, when the maximum distance is 50m, the maximum absolute error of pitch angle is less than 1.3°, the average absolute error is less than 0.32°, the maximum absolute error of yaw angle is less than 1.5°, and the average absolute error is less than 0.4°. The spatial position offset of 96% of the data points is less than 0.25m, which has achieved excellent results. Flight test experiments show that this method is feasible, the model is universal and has a wide application prospect.