COMBINATION OF PROGRESSIVE GROSS ERROR DETECTION WITH ROBUST
ESTIMATION ON A QUASI ON LINE ERROR DETECTION IN AEROTRIANGULATION
Joao Bosco Lugnani
Universidade Federal do Parana
Curitiba - Parana - Brasil
Edson Aparecido Mitishita
Aerodata S/A.-Engenharia de Aerolevantamentos
Curitiba - Parana - Brasil
ABSTRACT
This paper presents the results of a research conducted by the
authors, seeking to improve reliability and efficiency in time and
economy, when simultaneous off-line block aerotriangulation is
performed.
Frequently the conditions in which the simultaneous block
aerotriangulations are performed, particulary in Brazil and probably
in many other countries, sum up a great number of sources of gross
errors that cause serious problems for the processing of the
simultaneous block adjustment. These errors increase the processing
time, cause doubts in the detection of gross errors, reduce the
precision and reliability of the results.
A set of algorithms and a system of programs were conceived and
implemented to detect gross errors, progressively, on a quasi on
line basis, as the measurements are conducted. The algorithm were
developed, implemented and tested for both, bundle block and
independent model aerotriangulation adjustment.
The results of the tests showed great imprevements in time of
processing, efficiency in the detection of gross errors and in
reliability. Robust estimation was applied in some steps of the
sequential detection to encrease its efficiency.
1. INTRODUCTION
It has been an area of attention of the Photogrammetric Community
the problem of reliability of the estimated parameters of
phototriangulation. Reliability is closely related to detection and
location of gross errors, that affects also the processing time and
cost.
Progress has been achieved in this area during the last few years.
Some methods like the Data Snooping [02], [03]: The Danish Method
[08], [07]; the Step-by-Step, [01], [06]; and the On-Line, [09];
have improved reliability. Some of them bring benefit also to the
saving and others bring heavy load of processing time. The
combination of methods has also been implemented and tested [06],
[04], with benefits for the area.
- 442 -