A BRIEF OVERVIEW OF THE CURRENT STATE, CHALLENGING ISSUES AND FUTURE DIRECTIONS OF POINT CLOUD REGISTRATION

Point cloud registration is the process of transforming multiple point clouds obtained at different locations of the same scene into a common coordinate system, forming an integrated dataset representing the scene surveyed. In addition to the typical target-based registration method, there are various registration methods that are based on using only the point cloud data captured (i.e. cloud-to-cloud methods). Until recently, cloud-to-cloud registration methods have generally adopted a coarse-to-fine optimisation process. The challenges and limitations inherent in this process have shaped the development of point cloud registration and the associated software tools over the past three decades. Based on the success of applying deep learning approaches to imagery data, numerous attempts at applying such approaches to point cloud datasets have received much attention. This study reviews and comment on recent developments in point cloud registration without using any targets and explores remaining issues, based on which recommendations on potential future studies in this topic are made.