Multiple scans of the same object acquired with 3D imaging system (e.g., laser scanner) in the same experimental conditions could provide valuable information about the instrument s performance (e.g., stability, existence of bias, measurement error). Geometrical primitive may be fitted to multiple datasets and the variances of the fitted object s parameters may serve as a measure of instrument s performance. We test this procedure on simulated data as well on the data acquired in a laboratory. Two different error functions (orthogonal and directional) are used to fit a sphere of known radius to the data. A spread of sphere centers fitted with the directional function to simulated data is in agreement with theoretically calculated variances of fitted centers. For sphere centers fitted to the data acquired in a laboratory, the variances do not agree with the spread. This fact is interpreted as an evidence of a non-zero bias in the recorded range data. The orthogonal fitting yields sphere centers in disagreement with theory both for simulated and laboratory datasets.
Citation: IEEE Transactions on Instrumentation and Measurement
Pub Type: Journals
Nonlinear Least Square fitting, 3D imaging systems, experimental noise, orthogonal error function, directional error function