ROBUST ESTIMATION IN PHOTOGRAMMETRY.

A description is given of the principles of robust computation methods which yield results uninfluenced by gross errors. The common characteristic of these methods is that the square of the residuals is not minimized, but another properly chosen function is minimized. The different classes of robust method are described and compared to other methods of blunder detection. The Danish method, which was developed by one of the authors and today is implemented in many commercial block adjustment programs, is outlined. The problems with this method are explained and guidelines are given for its proper use. Finally, practical results with this method are summarized.