光達系統為一新興點位三維幾何資訊收集利器,光達掃瞄點位為一離散型資料,必須藉助資料處理實施特徵萃取或分類,方能供後續使用。不論空載光達或地面光達其點雲資料數量皆十分龐大,隱含在其中的空間幾何特徵需經由有效的萃取演算法及適當的電腦計算效能方可順遂地完成特徵萃取之任務。 本研究使用三維網格框架置入點雲資料,藉網格編號界定點雲間位相關係,並以區域成長法搜尋近似平面參數,其後經由迭代式霍夫轉換萃取較精確之平面特徵。其平面特徵可進而衍生交線、交點特徵。經過實驗顯示,結合三維網格及迭代式霍夫轉換萃取特徵之結構化處理的確可行,並藉由誤差傳播可解算成功萃取之平面特徵、線特徵與交點特徵精度,提供特徵物於物空間之位置、幾何及理論精度供後續使用。
LiDAR systems have recently emerged as efficient tools for collecting geo-spatial information. Due to the discrete nature of point cloud, regardless of ground-based or airborne LiDAR systems, series of processes via designed algorithms and aids from the computer power on the data must be imposed before any features or information can be revealed. This research employs 3-D grid structure well addressing point cloud into 3-D topology, regional growing for hypothesizing planes, and iterated Hough Transform for refining the plane-features. Line features and point features can then be derived based on the previous solution on plane extraction. The experimental results show that the proposed structuralization scheme is applicable and the extracted features are to be satisfactorily utilized for relevant applications.