Abstract
Abstract. The optimal network design problem has been well addressed in geodesy and photogrammetry but has not received the same attention for terrestrial laser scanner (TLS) networks. The goal of this research is to develop a complete design system that can automatically provide an optimal plan for high-accuracy, large-volume scanning networks. The aim in this paper is to use three heuristic optimization methods, simulated annealing (SA), genetic algorithm (GA) and particle swarm optimization (PSO), to solve the first-order design (FOD) problem for a small-volume indoor network and make a comparison of their performances. The room is simplified as discretized wall segments and possible viewpoints. Each possible viewpoint is evaluated with a score table representing the wall segments visible from each viewpoint based on scanning geometry constraints. The goal is to find a minimum number of viewpoints that can obtain complete coverage of all wall segments with a minimal sum of incidence angles. The different methods have been implemented and compared in terms of the quality of the solutions, runtime and repeatability. The experiment environment was simulated from a room located on University of Calgary campus where multiple scans are required due to occlusions from interior walls. The results obtained in this research show that PSO and GA provide similar solutions while SA doesn’t guarantee an optimal solution within limited iterations. Overall, GA is considered as the best choice for this problem based on its capability of providing an optimal solution and fewer parameters to tune.
Highlights
Unlike methods that only capture specific individual points at a time, e.g., a total station or GPS, light detection and ranging (LiDAR) systems measure large amounts of 3D points with very high acquisition speed
This paper focuses on the comparison of three heuristic methods in the first-order design (FOD) of indoor terrestrial laser scanner (TLS) network optimization, whose principles will be introduced below
The greedy method solution shows that a sub-optimal plan with a minimum of 5 viewpoints for this case can be obtained with no iteration, and the impact of being away from the optimum will increase in case of more complex scenes
Summary
Unlike methods that only capture specific individual points at a time, e.g., a total station or GPS, light detection and ranging (LiDAR) systems measure large amounts of 3D points with very high acquisition speed. TLS quickly captures rich detail of an entire scene like a camera taking a 360° photo but with an accurate 3D position for every pixel. It determines the object position based on the time-of-flight or phase-shift between the laser beam emitted to the object and the corresponding reflected signal. Since the objects to be scanned are either large (e.g., a very tall building) or occluded/self-occluded (e.g., a complex industrial site), a scanning network consisting of multiple scan locations is usually required to provide complete coverage of the object, which is the focus of this paper
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
