以四旋翼UAS酬載熱感測器取像製作數值表面溫度模型

Generation of Digital Surface Temperature Model from Images Collected by Thermal Sensor on Quadcopter UAS

熱感測器能感測可見光感測器無法取得的訊息，因此若能透過熱感測器進行環境監測，便能得到一般可見光感測器無法獲取的資料。本研究以四旋翼UAS酬載熱感測器得到局部區域地面高解析度之熱像並製作數值表面溫度模型（Digital Surface Temperature Model, DSTM）供環境監測之用。而一般環境監測區，如火山地區，不易布設控制點，因此本研究除於UAS上酬載熱感測器之外，並搭載GNSS接收模組，探討GNSS觀測量輔助熱像定位定向之可行性，研究以國立政治大學旁的指南溪為實驗區，使用AI-RIDER YJ-1000-HC四旋翼UAS酬載FLIR Tau 640熱感測器以及Trimble BD970 GNSS接收模組，透過GNSS動態後處理資料，取得取像時對應的GNSS觀測量，再搭配少量地面控制點執行光束法平差輔助熱像定位定向，過程中透過三焦張量約制剔除自動匹配之誤匹配連結點。實驗結果顯示，所產製之數值表面模型（Digital Surface Model, DSM）於不易變動區，經現有資料檢核均在±1m之間，產製出地面解析度11cm的DSM與正射熱像，且正射熱像平面精度為47cm。雖然DSM和正射熱像精度無法符合一般常規的測量規範，但成果仍可證明熱像直接產製DSM以及正射熱像之可行性，最後生成數值表面溫度模型的成果可供後續環境監測研究使用。

Thermal infrared camera can sense some important information that general optical digital cameras cannot do for the environment monitoring. In this study, the Quadcopter UAS for thermal image collection applied to environment monitoring was developed. FIIR Tau 640 thermal infrared sensor was used in this thermal image collection system. Additionally, Trimble BD970 GNSS OEM board was also carried on the Quadcopter UAS to collect dual-frequency GNSS observations for determining the flying trajectory by Post-processed kinematic (PPK) technique to support the positioning and orientating of collected thermal images by bundle adjustment. Meanwhile the trifocal tensor constraint was used to delete wrong matching tie images points before performing bundle adjustment for positioning and orientation of thermal images. The test area was located at Zhinan River nearby National CengChi University, Taipei, Taiwan. From the test, the differences between produced DSM and existing DSM data were ±1 m on stable ground surface. Both the resolution of produced DSM and thermal orthoimages were 11 cm and the accuracy of thermal orthoimages could reach 47 cm. The accuracy of thermal orthoimages might not comply with a normal surveying standard, but it proved the possibility of DSM and orthorectifed thermal images generated from thermal images directly. Additionally, the produced Digital Surface Temperature Model (DSTM) could be used for environment monitoring in the future.