Region Segmentation Method of Mobile Robot Based on ORB Feature in Dynamic Scene

In the traditional mobile robot SLAM system, the requirements for the scene are generally static, because the dynamic target will have an impact on the mapping process, causing the system to calculate the position of the camera incorrectly, resulting in a decrease in the accuracy of the navigation process, leaving a series of non-static noise caused by dynamic objects, thus causing the system to record the scene incorrectly. In order to reduce the impact of dynamic objects on SLAM system, this paper proposes a dynamic object region segmentation method based on ORB algorithm, which aims to segment the dynamic region from the image before the image is input into slam system, leaving the static region, and then perform DMatch matching for the static region to achieve the effect of reducing the interference of dynamic objects.