Motion estimation method based on image features and three-dimensional information of three-dimensional visual system

Abstract

The invention discloses a motion estimation method based on image features and three-dimensional information of a three-dimensional visual system. The method includes the first step of obtaining two-dimensional image information and space three-dimensional information corresponding to two-dimensional image pixels in real time on the basis of the three-dimensional visual system, the second step of controlling collection intervals between adjacent frames through the adoption of an adjacent frame sampling interval self-adaptive regulation method and automatically regulating the sampling intervals of the t+1 moment relative to the t moment according to the dot pair number of the effective matching image features between the frame at the t-1moment and the frame at the t moment, and the third step of synthesizing two-dimensional scene image features and three-dimensional point cloud information and carrying out three-dimensional motion estimation according to the matching point pair number of the current two-dimensional image features of the adjacent frames to obtain high-precision three-dimensional motion estimation. Through the method, on the basis of satisfying accuracy and reliability of the three-dimensional estimation, the calculated amount is effectively reduced, and the method has the advantages of being less in calculated amount, small in limitation to an application scene and the like.

Description

technical field [0001] The invention relates to a motion estimation method based on image features and three-dimensional information of a three-dimensional vision system. Background technique [0002] High-precision and high-reliability 3D motion estimation is one of the research difficulties and hotspots in the field of robotics. Its goal is to calculate the change of the robot's 3D space position at two adjacent moments in real time. The core content of research such as tracking. The traditional inertial navigation system has a simple principle and is widely used in motion estimation, but it has problems such as drift and error accumulation, and its corresponding motion estimation has low accuracy and poor reliability. Compared with inertial navigation systems, vision-based motion estimation does not have the problem of physical drift and has higher stability. At present, motion estimation based on monocular vision cannot obtain the scale information of three-dimensional...

AI Technical Summary

Problems solved by technology

However, the defect of this type of method is that it relies heavily on the selection of feature points. First, the mismatching point set in the feature point set will have a serious impact on the initial value of the 3D estimation; secondly, the algorithm is only effective for scenes with many image feature points. If the scene features When the point set is so sparse that it is impossible to obtain a correct and reliable initial estimate, it will directly lead to the failure of the algorithm, so that the three-dimensional relative estimation cannot be obtained

Images