Quantification of edge loss of laser scanned data at spatial discontinuities

Laser scanning is a promising geometric data collection tool for construction, facility, and infrastructure management due to its fast sampling rate (tens of thousands of measurements per second) and millimeter-level accuracy. However, laser scanned data contains inaccurate data points at spatial discontinuities (object edges). These inaccurate points, known as mixed-pixels, are commonly removed from the data prior to geometric modeling or other downstream processes. The removal of points at the edges of objects introduces error in the geometry of the objects, and object dimensions extracted from the data, such as widths and heights, are usually smaller than the actual values. In many cases, these losses due to removal of points at edges can exceed measurement accuracy tolerances specified in inspection manuals. This paper proposes a model for estimating edge loss in laser scanned data by considering the impacts of various factors, such as scanning distance, density of data and incidence angle on the edge loss. Results from a series of controlled experiments showed that the developed model successfully predicted edge losses in most test cases. Evaluation results using data collected from job sites showed that this model reduced the measurement error due to edge loss by an average of 80% for dense point clouds collected by an amplitude modulated continuous wave (AMCW) scanner, and 38% for relatively sparse point clouds collected by a pulsed time of flight (PTOF) scanner. By adding the estimated edge losses back into the raw dimensional measurements using the developed model, it is possible to significantly improve the accuracy of related measurements and hence improve the accuracy of the geometric information extracted from laser scanned data.