An Improved Monocular ORB-SLAM Method

Abstract

An improved algorithm is proposed for the practical application of monocular vision ORB-SLAM algorithm. When the image FAST feature points are detected, the adaptive gray level threshold is designed to ensure the validity of the feature points detection when the image contrast is decreased due to the illumination changes in the environment, and the robustness of the SLAM algorithm is improved. Initialization of the environment for the plane of the situation, the design of the third frame image verification method to solve the problem of multiple solution affine model. Design hardware platform to validate the proposed method for improving monocular vision ORB-SLAM. The result show that the proposed method is practical and feasible, and it improves the robustness and self-adaptability of the ORB-SLAM algorithm, and has high practical value.