DETECTION OF DISPLACING POSITIONS AT A SLOPE BY VISION METROLOGY − APPLICATION OF THE DATA SNOOPING TECHNIQUE

Abstract

The Baarda's data snooping is applied to detection of early deformations of a slope by vision metrology. The data snooping is a method to detect gross errors by adding an error term to each observation equation, and if adjustment is suddenly improved at a point, the observation is judged to include a gross error. Phenomenally a deformation occurrence between two epochs of time corresponds to a gross error in observations. Equations for a hypothesis test are derived from a set of photogrammetric observation equations. The validity of the testing procedure is proved using a slope model of 1m×0.5m with 37 retro-targets, in which some points are shifted by known amount between two epochs.