INTERGEO Agenda - english

Deformation analysis using terrestrial laser scanners (TLS) becomes increasingly popular. More and more methods to determine even the smallest movements in TLS point clouds appear in current research. However, such tasks require precise knowledge of the underlying variances of the observations. The lack of information or the use of incorrect stochastic information leads to parameters without meaningful significance. We propose a method to determine the angular variances of a TLS empirically. Specifically, the Z+F IMAGER ® 5016A is used in the development process. We are able to determine both horizontal and vertical angular variances of given TLS. In combination with the well known intensity-based range variance model, this is the first time that the variances of all three observables of a TLS are determined empirically. Unlike existing approaches that mainly use target center estimation (TCE) based approaches, our method realizes the angular variance in the intensity observation. This allows the determination of the variance of an individual angular observation, rather than smoothing over multiple observations as is done in TCE. An in-situ quantification and reduction of the intensity measurement variance allows the decomposition of noise sources until the angular contribution is separated. Determined angular variances can be modeled with a functional relationship in dependence on the respective rotational speed. Both horizontal and vertical angular standard deviations, the square-root of the variances, are significantly smaller than the manufacturer’s claim of up to 14.4”. Determined horizontal angular standard deviations range from ≈0.7” to ≈2”, with higher rotational speeds leading to lower standard deviations. Vertical angular standard deviations show a range from ≈1” to ≈3” with higher rotational speeds leading to higher vertical angular standard deviations. In addition to the Z+F IMAGER ® 5016A, two TLS with identical hardware specifications but differences in firmware and age were analyzed. The results show similarities in determined angular variances for the same hardware and differences for scanners of different age and servicing history.