Positioning and orientation method based on multi-aperture imaging

Abstract

The invention discloses a positioning and orientation method based on multi-aperture imaging, and the method employs at least three cameras to form a multi-aperture imaging system, and all cameras have no common imaging area, which is equivalent to large-view-field and high-resolution joint imaging. Establishing a two-dimensional to three-dimensional corresponding set; performing equivalent monocular pose estimation based on pre-calibrated camera internal parameters and inter-camera external parameters, and solving a PnP problem to obtain the position and pose initial values of the multi-aperture imaging system; and constructing an observation model based on the multi-aperture imaging relationship, and optimizing the position and attitude initial values of the multi-aperture imaging system to obtain the high-precision position and attitude of the multi-aperture imaging system. The method is applied to the technical field of photogrammetry and computer vision, multi-aperture imaging is equivalent to ultra-large-view-field-range ultra-high-resolution imaging, the observation model based on the multi-aperture imaging relation is established, high-precision positions and postures are obtained through optimization, hardware limitation is broken through, and extremely-high-precision positioning and orientation are achieved.

Description

technical field [0001] The invention relates to the technical fields of photogrammetry and computer vision, in particular to a positioning and orientation method based on multi-aperture imaging. Background technique [0002] At present, positioning and orientation technologies mainly include inertial navigation and visual navigation. Although inertial navigation has good autonomous performance and does not rely on external information, high-precision inertial navigation is heavy, bulky, expensive, and limited by the mechanism, the error will diverge over time. Visual navigation, free of time drift and error accumulation, is another navigation method that has attracted much attention. There are many existing visual navigation methods. Generally, monocular, binocular or depth cameras are used to solve the n-point perspective problem (PnP) to achieve positioning and orientation, but the measurement accuracy still needs to be improved. High-precision positioning and orientatio...

AI Technical Summary

Problems solved by technology

[0003] High-precision positioning and orientation requires the imaging system to have both ultra-high object plane resolution and a large field of view. Under the existing hardware conditions, a single camera cannot meet the requirements of both aspects at the same time.

Images