Linear Structure Recognition Algorithm Based on UAV Color Point Cloud

Le résumé fourni par la source

Accurate identification of linear structures is crucial for terrain feature recognition. Traditional identification methods rely on manual observation, which are time-consuming, labor-intensive, and susceptible to subjective bias. Therefore, developing an efficient, accurate, and safe method for linear structure recognition holds significant practical importance and application value. This study proposes a linear structure recognition algorithm based on colored point clouds. The algorithm fully leverages high-quality 3D colored point cloud data acquired by UAVs, integrating geometric features and color information to achieve efficient and precise identification of linear structures. Methodologically, the approach begins with preprocessing of the colored point cloud and applies DBSCAN clustering based on color features to extract potential structural regions. Subsequently, local curvature and normal vector information are used to filter the initially detected structural areas. Finally, a region growing algorithm is employed to optimize the preliminary detection results, further enhancing the accuracy and completeness of the recognition.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.