Scan Context

Scan Context is global descriptor LiDAR Point Cloud. And it is designed for a sparse and noisy point cloud acquired in outdoor environment. (But recently, there is scan context++ which is robust to rotation and lateral variations in urban environments.) So it is a loop-closed detection / place recognition algorithm when you only have point cloud (geometry informations). This is a flow of Scan Context. 3D Point cloud is transformed to Scan Context(2D) by ring and sector. Then, nearest neighbor search process is needed because of angle differences. Although pose is same, as angle of scans, scan ..........

원문링크 : Scan Context