Calculate Error

Feature Description

The Calculate Error feature computes error information between registration control points and reference control points using the selected algorithm. Smaller errors indicate better registration results. Users can utilize registration info to determine whether to reselect control points for improved alignment.

The error units correspond to the current dataset's planar coordinate system. Typically, registration accuracy should be less than 0.5 pixel. For imagery with 30-meter resolution, the total RMSE should be under 15 meters.

Different projects may have varying accuracy requirements. When certain control points' RMSE exceeds acceptable thresholds, deleting or re-editing these points can reduce overall RMSE and enhance registration precision.

Feature Entry

  • Register tab->Operation group->Calculate Error button;

Steps

  1. Select a registration algorithm from Register tab->Operation group. Four algorithms are available: Linear Registration, Quadratic Polynomial Registration, Rectangular Registration, and Offset Registration. Choose based on actual requirements, noting different algorithms require varying numbers of control points. For algorithm details, see Registration Algorithm Introduction.
  2. Click Calculate Error in Operation group under Register tab to initiate computation. The control point list displays individual errors including X Residual, Y Residual, and RMSE. The status bar in registration window shows Total RMSE. For error interpretation, refer to Registration Error Overview.
 Tip:
  • Right-click any position in control point list and select Export Registration Info from context menu to save all control points as registration file (*.druf). Saved files can be imported for future use. For export operations, see: Import and Export Registration Files.

Related Topics

Step 1: New Registration

Step 2: Add Control Points

Step 4: Perform Registration

Import and Export Registration Files