Structure from Motion and Visual-Inertial Simultaneous Localization and Mapping for Three-Dimensional Reconstruction

The development of Structure from Motion (SfM) and visual-inertial Simultaneous Localization and Mapping (viSLAM) is gaining attraction. As a result, researchers figure out many methods and multiple versions of methods to solve the problem under various circumstances or within different cameras. SfM and SLAM are on two tracks but have a connection to the essence. The paper started by introducing the basic principles and methods of SfM and viSLAM, covering basic techniques and improvements to promote efficiency and stability when the input information varies from sequential to with depth information. This paper serves as a description and suggestion for whom wants to use SfM or viSLAM as three-dimensional reconstruction tools. This paper summarizes the focus of diverse methods and lists their outcomes to show the alteration or improvement. Then the pros and cons of methods are aggregated to give a specific comparison between reconstruction methods.