A UAV positioning system and method based on multi-view geometry and beam adjustment

Abstract

The invention belongs to the field of aerial photogrammetry and particularly relates to an unmanned aerial vehicle positioning system and method based on multi-view geometry and bundle adjustment. According to the method, a photo selection module selects two photos to perform feature extraction and matching; relative rotation and translation of two cameras are calculated according to matching feature points; whether to add the image pair into a target positioning calculation module is judged according to the matching number of the feature points and a base line distance between the two cameras; a multi-view geometry algorithm is adopted to calculate three-dimensional coordinates of relative external participating feature points of the cameras, the calculation result is used as an initial value of a nonlinear adjustment system to perform adjustment, and three-dimensional coordinates of the relative external participating feature points of the cameras after adjustment are obtained; a rotation-translation matrix of a relative coordinate system and a longitude and latitude coordinate system is calculated according to actually-measured gps coordinates of the photos; and last, longitudes and latitudes of the external participating feature points of the cameras under an absolute coordinate system are obtained, and the longitude and latitude of a target point are obtained according to difference values of the longitudes and latitudes of the four feature points closest to the target point.

Description

technical field [0001] The invention mainly belongs to the field of aerial photogrammetry, and in particular relates to an unmanned aerial vehicle positioning system and method based on multi-view geometry and beam adjustment. Background technique [0002] At present, for UAV target positioning, the initial value is mainly calculated by the forward intersection algorithm, and the nonlinear adjustment algorithm is used to obtain the high-precision target position. The forward intersection algorithm is an algorithm that calculates the three-dimensional coordinates of the target by tracking the target on a continuous image sequence and according to the external parameters of the camera provided by the sensor; Eliminate algorithms that result in more accurate measurements. The whole target positioning system is composed of three modules: calculation of initial value of forward intersection, solution of optimal solution of error equation and iterative optimization. The error eq...

AI Technical Summary

Problems solved by technology

However, the application of visual information to the initial value calculation needs to further solve the following problems: (1) how to obtain the internal and external parameters of the drone and the initial value of the target point coordinates using only the image sequence; (2) how to achieve fast target positioning; (3) how to eliminate Data with large errors, thus increasing the anti-interference ability of the algorithm and other issues

Images