空載光達作業流程及品質管理之研究

Abstract

利用空載光達獲取資料已逐漸成為生產數值高程模型主要作業方法之ㄧ，本研究針對其作業過程及產品的品質管理進行探討。 一般檢核空載光達資料的方式為比較平面及高程精度，此時比較的數據已為最終成果。若不符要求，則需重新產製，不合經濟效益。故本研究以「以流程保證品質」的概念探討查核與品管。整理現有之作業流程為一簡單明瞭之流程，並針對一般應用目的設計所需之查核項目與時機，然而各項目及參考數值可依使用目的不同加以調整。研究中針對幾項影響重大的因素進行進一步探討。包含點雲密度、點雲處理與編修及高程精度檢核。 經由實驗發現原始點雲數量不足，連帶影響地面點數量不足造成自動分類錯誤與DEM的不合理，而藉由適當的人工編修可有效改善產品品質，但仍無法保證完全正確，最後進行正高化算所使用之大地起伏模式之精度會影響到正高系統之DEM，其中以經過水準點修正之模式較為理想。

Airborne LiDAR has become one of the primary methods used for DEM generation. This study investigates the production process and quality management scheme for DEM generation from airborne LiDAR data. The quality of airborne LiDAR produced by DEM is currently evaluated by identifying the planimetric and height position accuracies with ground checkpoints. Because there are only a limited number of checkpoints, and the process is usually conducted during the final stage, a step-by-step quality management scheme during the flow of DEM production would be beneficial. In this study, the production process is investigated first. Subsequently, a quality check scheme is designed. Based on different application requirements, the standard of the threshold values may be different. However, reference values are provided in this study. Several important issues, including point density, filtering, and manual editing, are studied together with the height accuracy validation. It has been determined that the distribution and density of ground points in the original point cloud greatly influence the DEM production. If the amount of points on the ground is not sufficient, manual editing may improve the overall data quality. However, the quality of the final product is still in doubt. The last step in the DEM production is the transformation from the ellipsoid height system into an orthometric height system. The performance of several currently available geoid models of Taiwan is also evaluated. It is found that models with differential leveling benchmarks included in the formation process perform better.