Image processing method and device

Abstract

The emdoiment of the application provides an image processing method and device. While locationing and environment sensing pricison are enhanced, waste of resources is reduced and cost is saved. The method comprises the following steps: obtaining a multi-frame image shot by a monocular camera; extracting multiple key frames of images from the multi-frame image according to position and posture relations of the multi-frame image shot by the camera; and calculating depth information of feature points of the multiple key frames of images by utilizing position and postures of the multiple key frames of images shot by the camera.

Description

[0001] Copyright statement [0002] The disclosure of this patent document contains material that is protected by copyright. This copyright belongs to the copyright owner. The copyright owner has no objection to the reproduction by anyone of the patent document or the patent disclosure as it exists in the official records and files of the Patent and Trademark Office. technical field [0003] The present application relates to the field of image processing, and more particularly, to an image processing method and device. Background technique [0004] With the development of modern technology, drones are used in more and more scenarios. To achieve automatic operation, drones need to have certain positioning capabilities and environmental awareness capabilities. [0005] At present, drones can use structured light cameras for positioning and environment perception. This kind of camera can directly obtain depth information, but this kind of camera performs poorly in strong li...

AI Technical Summary

Problems solved by technology

[0005] At present, drones can use structured light cameras for positioning and environment perception. This kind of camera can directly obtain depth information, but this kind of camera performs poorly in strong light environments, is susceptible to interference, and is relatively expensive.

[0006] Another commonly used is the dual / multi-eye vision module for positioning and environment perception, which consists of two or more cameras separated by a certain distance, and calculates the three-dimensional depth information by relying on the difference between the photos taken by the two cameras at the same time. The rotation and position between multi-eye cameras directly affect the effective range and accuracy of its measurement, so the dual / multi-eye vision module has relatively high structural requirements

Images