透過您的圖書館登入
IP:18.227.48.131
  • 學位論文

數值地形模型錯誤偵測及改正

Error Detection and Correction for Digital Terrain Models

指導教授 : 趙鍵哲

摘要


數值地形模型(Digital Terrain Models,DTMs)為記錄三維地面特徵與空間分佈的幾何資料,應用層面相當廣泛,是土地利用上的一項重要輔助資訊。而目前無論是經由航測作業或是其它自動化方式所生產之數值地形模型,大多仍需再倚賴人工進行資料檢核與修正的處理,但如此耗時費力的工作,多容易因作業疲累而致使成果品質不甚理想。 本研究探討以影像處理方式,建立一套DTMs品質控制(Quality Control,QC)的自動化檢核機制來偵測與改正錯誤。主要具體作法架構於下述假設: 以數值地形模型及正確無誤的影像和方位參數所產製的正射影像,品質與數值地形模型資料息息相關,即若數值地形模型上含有錯誤點位,將造成正射像對於應位置上的灰階及幾何資訊錯位,故據此理論採用區域式匹配法比對重疊區內的正射像對,透過灰階變異來偵測錯誤高程,評估數值地形模型之品質,並迭代進行自動化改正錯誤的處理作業。 此外,本研究再依據自然地形起伏的連續特性,設計另一種高程檢查工具輔助偵測錯誤。最後,綜合所有偵測工具之檢查成果來評估DTMs資料的品質,做為引導後續進行人工或其他應用時的參考資訊,以達到提昇作業效能的目的。 實驗結果顯示:錯誤偵測作業,可在濾除五成以上資料檢查工作量的同時,偵測得五成以上的錯誤量;錯誤改正作業,對於提昇資料品質的效果尚較不顯著,平均僅能自動改正兩成左右的錯誤量。雖然目前測試的成果仍未臻理想,但本演算法的基本完整架構已可做為後續研究工作的參考。

並列摘要


DTMs is an important land-use information, which provides geometric data of the terrain and can be realized in versatile applications. Producing the DTMs by the photogrammetric means or other automatic methods always needs visual inspection to evaluate the quality and correct the errors. However, such a labor exhausted and time consuming work makes mistakes easily. In this study, the author investigates the DTMs quality control system by employing image processing to automated check and improvement of DTMs errors. The fundamental point is rooted in the relationship between DTMs and the produced orthoimagery by the accurate orientation and steady imaging, the geometric and radiometric deviation of conjugate points in the orthoimage would imply the erroneous DTMs. Therefore, the result of image matching of two overlapping orthoimages would lead to the DTMs evaluation and automatic errors correction. Furthermore, according to the basic natural attribute, such as the terrain relief changes rather continuously, we design another auxiliary checking method to detect errors. In the end, the author integrates all the checking results to provide a reference information for DTMs quality assessment and for manual inspections to increase work efficiency. The experiment results show that the error detection can find out over 50% blunders and decrease over 50% inspections simultaneously. The error correction is not obvious and only can correct about 20% errors automatically. Although the achievement of the present test is not perfect enough, but it is possible to take the concept of the method as the template of the following research work.

參考文獻


楊善智,2007。數值高程模型(DEM)之品質評估,國立成功大學測量及空間資訊學系碩士論文。
楊榮華,2002。應用灰階共變異矩陣之多重指標於瑕疵檢測,元智大學工業工程與管理學系碩士論文,pp.8 ~ 27。
Georgopoulus, A. and D. Skarlatos, 2003. A Novel Method for Automating the Checking and Correction of Digital Elevation Models Using Orthophotographs, Photogrammetric Record, 18(102): pp.156-163.
Höhle, J.K. and M. Potucková, 2005. Automated Quality Control for Orthoimages and DEMs, Photogrammetric Engineering & Remote Sensing, 71: pp.81-87.
Manzer, G. , 1996. Avoiding Digital Orthophoto Problems, Digital Photogrammetry: An addendum to the Manual of Photogrammetry, pp.158-162.

被引用紀錄


陳芷儀(2011)。以數位航測探討南仁山地區森林孔隙動態變化〔碩士論文,國立屏東科技大學〕。華藝線上圖書館。https://doi.org/10.6346/NPUST.2011.00210

延伸閱讀