Visual slam initialization method, system and device in dynamic environment

Abstract

The invention belongs to the fields of robots, unmanned driving, and AR vision, and specifically relates to a visual SLAM initialization method, system, and device in a dynamic environment, aiming to solve the problem that SLAM lacks extraction of static feature points in a dynamic environment. The system method includes obtaining two frames of images with parallax; obtaining the matching feature points of the previous frame of images; dividing the previous frame of images into equal parts to obtain multiple image blocks, and using the image blocks containing matching feature points greater than the set threshold as grids model, and obtain the interior point and centroid of the lattice model; calculate the coupling degree of the lattice model through the interior point, select the lattice model based on the preset coupling degree threshold to construct the corresponding lattice model set; calculate the distribution variance of each lattice model set based on the centroid , select the set with the largest value to construct a static feature set; triangulate the static feature set, and initialize SLAM through nonlinear optimization. The invention can extract enough static feature points.

Description

technical field [0001] The invention belongs to the fields of robots, unmanned driving, and AR vision, and in particular relates to a visual SLAM initialization method, system, and device in a dynamic environment. Background technique [0002] Visual SLAM is a system engineering. In the classic visual SLAM framework, it is often assumed that the environment is a static environment, or the static features in the environment occupy the majority. Although more mature visual SLAM systems have emerged, such as ORB-SLAM, DSO, LSD, etc., due to the assumption of a static environment in the SLAM system, its robustness in moving scenes is extremely poor. Whether it is a direct method or an indirect method, the visual SLAM system lacks the interference of moving objects, and in the process of SLAM operation, the solution to the camera pose is completely dependent on the RANSAC method. If the static feature points are not in the entire image. If it occupies the majority, or the propor...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the above-mentioned problems in the prior art, that is, in order to solve the problem that the existing visual SLAM system lacks the extraction of static feature points in a dynamic environment, resulting in the inability to solve the problem of the pose of the camera in the SLAM system in a real-time and robust manner, the first method of the present invention On the one hand, a visual SLAM initialization method in a dynamic environment is proposed, which includes:

Images