Camera pose estimation method oriented to RGBD (Red, Green and Blue-Depth) data stream

Abstract

The invention relates to a camera pose estimation method oriented to an RGBD (Red, Green and Blue-Depth) data stream. The method comprises the construction of a three-dimensional point cloud mapping graph and the registration of a three-dimensional point cloud mapping graph. The method comprises the following steps that: firstly, carrying preprocessing operations including filtering and the like on depth data, and then, constructing the three-dimensional point cloud mapping graph by the preprocessed depth image according to the internal parameter of a camera; according to the camera pose of a previous frame, adopting a projection mapping algorithm to carry out corresponding point matching on the two frames of three-dimensional point cloud mapping graphs, carrying out energy optimization on a point-to-plane distance error function, and establishing the six parameters of a transformation matrix between two frames of point clouds; and according to the camera pose of the previous frame, estimating the camera pose of a current frame, and carrying out iteration on above process optimization to obtain a final camera pose. The method can be operated in real time on existing mobile equipment, and a good result can be obtained when the resolution of input data is low.

Description

technical field [0001] The invention belongs to the field of computer vision and computer graphics image processing, in particular to a camera real-time pose estimation method for RGBD data streams, the method can be used when the resolution of input data is low, depth data has holes and noise, equipment operation Estimating the camera pose in real time with low capability is of great significance for the research of SLAM and real-time 3D reconstruction technology. Background technique [0002] With the popularity of depth sensors and the development of 3D reconstruction technology, research on 3D model reconstruction based on depth data is emerging in recent years. Compared with traditional 3D reconstruction based on RGB data, depth data provides 3D information of the scene, which greatly improves the feasibility and accuracy of 3D reconstruction. The estimation of camera pose plays a vital role in related application scenarios such as 3D reconstruction. [0003] At prese...

AI Technical Summary

Problems solved by technology

[0004] Some of the above camera pose estimation methods are time-consuming, such as the extraction of feature points, the optimization of energy equations and other steps, and some have high requirements on the quality of input data, such as consistent illumination and high resolution, etc., which are usually difficult to move Meet real-time requirements on platforms with low computing efficiency such as equipment

Images