Terrestrial laser scanning observation network planning for large-scale building façades integrating local and global optimization methods

Terrestrial laser scanning observation network (TLS-ON) planning for large-scale building façades faces a challenge in balancing the data coverage and deployment cost. Local optimization methods are computationally efficient, but they often generate redundant scan positions. However, global optimization methods suffer from a high computational cost and a reliance on a high-quality initial population. In this paper, we propose a novel method integrating local and global optimization strategies for TLS-ON planning with high efficiency and quality. First, a greedy algorithm generates a locally optimal solution. A Gaussian kernel function is then developed to construct a high-quality and diverse initial population, based on this local solution. Finally, the constrained non-dominated sorting genetic algorithm II (NSGA-II) combined with the Pareto-Edgeworth-Grierson (PEG) method is applied to obtain a globally optimal network. Experimental validation at a large-scale campus site in China demonstrates that the proposed method reduces scan positions by 13.8% and improves coverage by 2.41%, compared to local optimization, while accelerating convergence by a maximum of 5.5 times over global optimization with random initialization. Real scanning cases confirm its practicality, achieving high coverage in field validation. This work offers an efficient solution for 3D data acquisition in applications such as building information modeling and urban digital twins.