Surfel 기반 깊이 이미지 업데이트를 통한 동적물체 제거 맵 생성 기법

Abstract

In this paper, we propose a novel algorithm for reconstructing a map with removed dynamic objects based on surfel. The proposed method removes the surfels corresponding to dynamic objects by comparing the depth from the sensor and the depth of the surfels from the map projected onto the camera frame and updating them. As a result of our algorithm, we were able to successfully reconstruct a static map in an indoor environment where a large number of people pass. We compared the raw point cloud converted from depth image, octomap[3], and the proposed method to validate the performance of our works.