Real-time three-dimensional reconstruction key frame determination method based on position and orientation changes

Abstract

The invention provides a real-time three-dimensional reconstruction key frame determination method based on position and orientation changes. The method is achieved through four steps of current frame positioning, positioning quality evaluation, view difference calculation and judgment on whether key frames are inserted or not. The method has the advantages that the key frames are high in quality, speed and stability, and the insertion of a wrong frame can be effectively prevented by judging the positioning quality of image frames; meanwhile, the view difference concept is put forward, it is ensured that each inserted key frame includes certain new image information through the corresponding view difference threshold values, redundancy between the key frames is effectively reduced, and burdens on map establishment are reduced.

Description

technical field [0001] The invention relates to the field of visual autonomous navigation of robots and unmanned aerial vehicles, in particular to a method for determining key frames of real-time three-dimensional reconstruction based on pose changes. Background technique [0002] Simultaneous Localization and Mapping (SLAM) in the field of robotics, also known as Concurrent Mapping and Localization (CML), was first proposed by Smith, Self and Cheeseman in 1988. Due to its important theoretical and application value, it is considered by many scholars as one of the key technologies to realize a truly fully autonomous mobile robot. After years of development, it has been applied in some robots. It is mainly divided into two branches: methods based on probabilistic models such as EKF-SLAM and Fast-SLAM, and methods based on multi-view geometry, such as Structure from Motion (SfM) in the field of machine vision. The latter can cope with larger scenes, faster movement, positioni...

AI Technical Summary

Problems solved by technology

[0008] 1. Without considering the image quality and positioning quality of the current key frame, it is easy to cause low-quality or even wrong frame insertion, which will bring serious consequences to the system;

[0009] 2. In some cases, such as when the camera shakes left and right in the same place, a large number of redundant key frames will be generated, which wastes the running time of the system

[0012] 1. Since the key frame selection method does not consider the field of view changes caused by camera rotation, key frames will not be added when the camera rotates more and pans less;

[0013] 2. It is fixedly required that the positioning quality of each key frame is very high before it is selected as a key frame, but it is impractical in actual situations. If you do not add key frames at key moments in time, it is easy to cause subsequent frames can not be positioned, resulting in system failure

Images