Year: 2026
Type: article
Abstract: Identifying the different elements from the information provided by sensors is a key factor for a successful navigation. In this sense, this paper aims to evaluate the binary segmentation of different reticular structures using a proposed algorithm over 3D spatial information in simulated outdoor environments. The information has been generated synthetically by simulation using outdoor environments with multiple lattice structures. For this purpose, this work proposes, first, a specifically designed algorithm which uses well-established techniques such as RANSAC or region growing to achieve a binary segmentation. Second, this ad hoc algorithm is evaluated against a wide range of different structures. The results of the algorithm were evaluated using established metrics, including precision, recall, and F1-score, as well as its computational time. The experiments showed the good segmentation performance of the algorithm for all evaluated structures. The proposed algorithm demonstrates robustness and simplicity in the binary segmentation of reticular structures of different sizes in outdoor environments. All the code as well as the data used in this work can be found open source at https://github.com/Urwik/arvc_ground_filter
Source: SN Computer Science
Institution: Universitat de Miguel Hernández d'Elx
Language: English
Subfield: Aerospace Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $3,190
OpenAlex ID: W7138359088