Cloud observation system and method based on multi-UAV and machine vision

Abstract

The invention provides a cloud observation system based on multi-UAVs and machine vision, including an observation unit, a positioning attitude unit, a visual image matching unit and a three-dimensional wind velocity field reconstruction unit of the cloud body, and the observation unit is coordinated by at least two UAVs Watching the cloud group to shoot video, the positioning and attitude unit accurately obtains and controls the distance and attitude angle of the camera equipped on the drone; the visual image matching unit uses the cloud shape captured by the drone obtained by the positioning and attitude unit to compare with other drones The multi-area spatial correlation coefficient of the red light image is used to find the highest correlation point at the pixel level to match; the cloud body three-dimensional wind velocity field reconstruction unit uses the pixel positions of the same cloud group in multiple drones at the same time, and the drones at this moment The positioning and attitude parameters of the cloud cluster are reconstructed using the three-dimensional multi-view projective geometry algorithm. The present invention solves the problems existing in various existing cloud observations through the system and method for three-dimensional observation of cloud bodies and reconstruction of the three-dimensional wind velocity field.

Description

technical field [0001] The present invention relates to many technical fields based on unmanned aerial vehicle and machine vision observation, in particular, relates to a cloud observation system and method based on multi-unmanned aerial vehicle and machine vision. Background technique [0002] Global climate change has become one of the research fields that attract the most attention from the scientific community and the public today. Current projections of future climate, especially the contribution of cloud radiative feedbacks, remain subject to enormous uncertainties. To fully understand the basic physical mechanism behind cloud radiative feedback, fine 3D cloud spatiotemporal data are needed to study convective thermodynamic processes. [0003] At present, there are many traditional cloud observation methods such as satellite optical observation, satellite CT technology, aircraft cloud penetration, Doppler radar and ground optical observation, but none of them can solv...

AI Technical Summary

Problems solved by technology

It solves the problems of low resolution, inability to obtain three-dimensional structure, and inability to track continuously in various existing cloud observation technologies

Images