On gross error detection methods for control points in image registration

The control point selection is a necessary step of image registration, which has an influence on the global efficiency of registration. Therefore, to ensure the quality of control points is an important issue. By using three error detection methods, such as data snooping, Tau test, and Danish robust estimation, this paper intends to explore the quality control issue of control points for image registration. Because the measurement error of control points has different levels of influences due to their variable geometric positions, a suitable method is required for the gross error removal of points. In the case study of the control point in image registration, Tau test is a more efficient method for detecting gross errors than data snooping and Danish robust estimation.