Curvature-Based Registration Accuracy Analysis

Point datasets have their unique coordinates.The rigid-body transformation process that moves point data with unique coordinates obtained for the same object to common coordinates and converts the overlapping parts into one is known as registration.Such registration may involve using multiple methods simultaneously, depending on the situation, and even when the same method is used, it may be repeated several times to satisfy the user-defined threshold and reduce registration errors.Analysis of the registration precision is important in this process.If this analysis is not clear or provides different results each time, the criteria for judging the precision become ambiguous.To solve this problem, a method is proposed for estimating the curvature from two datasets on common coordinates after registration, obtaining curvature sample points using a sampling process, and then calculating their distribution to determine whether the registration was successful.