Mansoor Sabzali*, Lloyd Pilgrim
To determine the robust self-calibration approach for Terrestrial Laser Scanner (TLS), the profound knowledge of Geodetic Network Design and Photogrammetry is required. For any photogrammetric tasks, three main predefined criteria - precision, correlation, and uncertainty of parameters - play more vital roles than other criteria. Geodetic network design is composed of four interrelated design orders: zero, first, second and third order of design to fulfil the mentioned criteria. Zero order design which is the core focus of this research reveals the correlation between estimated parameters in self-calibration of TLS. In other words, three types of the parameters - calibration parameters (CP), exterior orientation parameters (EOP), and object points (OP) – as unknowns must be solved through bundle block adjustment (BBA) of TLS self-calibration. The current parametrization of exterior orientation used for TLS calibration, unlike camera calibration, is limited to collinearity conditions from one scan station. According to established concepts in computer vision, by adding the constraints of relative orientation (RO) to bundle block adjustment, the estimations of unknowns will be in higher quality and potentially lower correlation. Therefore, the application of this principle must determine more precise and lower correlated parameters for TLS self-calibration. This research will evaluate the correlation of TLS self-calibration parameters between collinearity and proposed coplanarity conditions and will identify the potential improvements in correlation and precision of the parameters with the aid of the new formulation. An experiment of selfcalibration was undertaken using Leica ScanStation P50 and implemented on MATLAB codes.
Keywords: collinearity, coplanarity, correlation, precision, self-calibration