車載光達系統於衛星訊號受到遮蔽時,由於慣性導航系統獨立估算載具位置與姿態之誤差將隨時間累積,使得點雲資料之誤差行為不僅具有累積性也與訊號遮蔽時段有關,因此須對點雲進行自適應分段改正。本研究採用路面標線特徵作為自適應分段改正之共軛點,由萃取之路面點雲產製強度影像,並利用影像增顯提升路面特徵之辨識度與對比度,再由影像共軛匹配技術獲取共軛特徵點並反投影回三維空間,以動態門檻之粗差偵測策略提升共軛匹配點之正確性,最後,進入自適應分段改正之流程。實驗成果顯示,所提出的方法可提供精確且足夠的共軛特徵,以利後續自適應分段改正,而多時期點雲間之誤差則可由數十公分之誤差下降至公分等級,能有效消除衛星訊號遮蔽造成之誤差。
Mobile LiDAR systems have been widely used in collecting 3D spatial information due to its high efficiency. However, its positioning quality relies on a good reception of GNSS signals. Therefore, point clouds positioning error accumulates with time when GNSS signal is obstructed. This study presents a fully automatic intensity-based multi-epoch mobile LiDAR point clouds matching and adjustment procedure. The experiment results indicated that the proposed frameworks can provide accurate and sufficient features for the adaptive time-variant adjustment and help to improve the positioning quality for the LiDAR point clouds acquired in a GNSS signal obstructed area. Consequently, the mobile LiDAR systems can be extended to a wider field of applications.