以空載光達校正場矩形平屋頂半自動評估點雲精度之程序研究

以空載光達校正場矩形平屋頂作為標的物執行點雲精度評估時，需先萃取平屋頂面上之點雲，為避免萃取成果受雜訊影響，本研究引入粗差偵測理論，發展最小一乘法[1]結合李德仁所發展以後驗變方估計原理導出的選擇權迭代法（稱李德仁法）[2]將非屋頂點視為粗差排除。研究中使用校正場6組點雲資料，經由測試提出半自動化點雲精度評估之程序。實驗結果顯示本研究所提之程序，其成果可避免因人而異，達一致性之點雲精度評估結果；而實驗結果亦顯示，本研究所提最小一乘法結合李德仁法的改良李德仁法，可將非屋頂點視為粗差排除，於含有大量牆面點雲的資料中，比原李德仁法提升約10%的偵測粗差數量。

關鍵字

點雲精度評估；最小二乘平面擬合；粗差偵測；最小一乘法；選擇權迭代法

並列摘要

This study attempts to establish a semi-automatic procedure for the accuracy assessment of point clouds from airborne LiDAR system. Least Absolute Deviation (LAD) [1] combined with the Iteration using selected weights presented by Li (Li method) [2] is developed to exclude the non-roof points that are regarded as gross errors and eliminate their influences. In this study, six datasets in calibration field were used to develop a semi-automatic procedure based on several tests. Even different operators employ this semi-automatic procedure, consistent results were obtained and the reliability can be achieved in this study. In additional, about 10% gross errors can be detected by our proposed improved Li method than original Li method.

並列關鍵字

point clouds accuracy assessment ； least squares plane fitting ； gross error detection ； least absolute deviation ； iteration with the selected weights

參考文獻

Boscovich, R. J., “De literraria expeditione per pontificiam synopsis amplioris operis, ac habentur plura eius ex exempla ria etiam sensorum impressa,” Bononiensi Scientiarum et Artium Inst. Atque Acad. Commen, No. 4, pp. 353-396 (1757).