半自動化建物模型與影像之套合是目前航空攝影測量主要的研究方向之一，其中模型與影像套合的演算法是其中關鍵所在。周宏達[2001]利用影像萃取點至模型投影邊線的距離為觀測函式，提出一套最小二乘法進行模型與影像套合的理論，提供一個模型與影像套合最佳化的方法。這套理論雖嚴謹可行，但過去的實驗數據顯示，有許多因素會影響套合的結果，甚至產生錯誤的套合。因此，本研究的主要目的是分析這些影響因素，並謀求對策減少套合的錯誤率。

模型與影像套合第一個影響因素是影像邊緣線像元萃取情形。本研究先針對不同的特徵萃取偵測元來分別進行套合，討論不同的邊線偵測元對套合結果的影響。接著為了降低在模型框線環域內非建物之萃取像元的干擾，利用特徵萃取所能得到的影像灰階之梯度強度及梯度方向等資訊，並處理成萃取像元的權，期望在最小二乘法套合時，可以不受非建物邊線像元的影響。最後決定出一個最佳的邊線偵測元與萃取點權值大小的組合，進行模型參數收斂範圍、影像重疊張數的實驗。而對於套合錯誤的情形，則提出兩個修正方法，分別是人工增加萃取點及模型參數的約制，使得模型參數可以順利解算。

參數式模型也是模型與影像套合的重點所在，愈多的模型代表可套合的建物愈多。本文針對實際影像增加兩個模型元件，分別是內縮三角柱及圓柱體。特別是圓柱體的套合，驗証了使用本文的方法仍然可對曲面元件進行套合。

Semi-automatic model-based building extraction is currently one of the major topics in the field of digital photogrammetry. How to design an optimal algorithm for model-image fitting is the key to this issue. Based on the previous study, a least-squares approach has been proposed for the optimal model-image fitting. The principle of this method is to minimize the distance between image edge pixels and the projected wire frame by adjusting the model parameters. This method is feasible in most cases. However, incorrect fitting results may happen due to some factors, such as the applied edge detection method, the weight of edge pixels and applied constraints. Therefore, the main purpose of this paper is to analyze the influences of these factors and improve the algorithm to reduce the probability of error fitting.

The firstly probed factor is the method used for edge detection. The results of model-image fitting using different edge detection methods are analyzed. In order to reduce the influence of the non-building edge pixels within the buffer of the projected model lines, we use the edge strength and direction of the edge pixels as the weights of the least-squares fitting. The experiment results that the best weighting method is to combine the factors of edge strength and direction. Furthermore, two strategies were used to avoid the error fitting. One is to add the building edge pixels artificially and the other is to constrain the model parameters more restrictively.
The number of building model is also important in model-image fitting. More the number of models, buildings can be model better in detail. In this paper, two new models are used for the model-image fitting. Especially, the fitting result of a cylinder model demonstrates that the fitting method can be applied successfully to a primitive with curved surface.

王正忠，「以近景攝影測量進行模型式建物重建」，國立成功大學測量工程研究所碩士論文，2002。

林文棋，「半自動化建物萃取之建物模型建置與操作」，國立成功大學測量工程學系碩士論文，2001。

林志交，「以基因演算法於模型影像套合計算之應用」，國立成功大學測量工程研究所碩士論文，2002。

周宏達，「以最小二乘法進行參數式模型與影像之最佳套合」，國立成功大學測量工程研究所碩士論文，2001。

Förstner, W., "A Framework for Low Level Feature Extraction", Computer Vision-ECCV '94, Vol. 2, Springer-Verlag, pp. 383-394, 1994.

Förstner, W., "3D-City Models: Automatic and Semi-automatic Acquisition Methods", Photogrammetric Week '99, Stuttgart, Vol. 47, pp. 291-304, 1999.

Fua, P., "Model-based Optimization: Accurate and Consistent Site Modeling", International Archives of Photogrammetry and Remote Sensing, Vol. 31, Part B3, pp. 222-233, 1996.

Grün, A., "Automation in Building Reconstruction", Photogrammetric Week '97, Herbert Wichmann-Verlag, Heidelberg, pp. 175-188, 1997.

Grün, A. and Dan, H., "TOBAGO-a Topology Building for the Automated Generation of Building Models", Automatic Extraction of Man-made Objects from Aerial and Space Images(II), at Monte Vertia, pp. 149-160, 1997.

Grün, A., and X. Wang, "Cyber City Moedler, A Toll for Interactive 3-D City Model Generation", Photogrammetric Week '99, Vol. XXXXⅦ, pp. 317-327, 1999.

Gülch, E., "Virtual Cities from Digital Imagery", Photogrammetric Record, 16(96), pp. 893-903, 2000.

Lang, F., Löcherbach, H. and Schickler, W., "A One-eye Stereo System for Semi-automatic 3D-building Extraction", Geomatics Info Magazine, pp. 1-4, 1995.

Lang, F. and Förstner, W., "3D-City Modeling with a Digital One-eye Stereo System", International Archives of Photogrammetry and Remote Sensing, Vienna, Vol. 31, pp. 415-420, 1996.

Rottenstener, F., "Semi-automatic Building Reconstruction Integrated in Strict Bundle Block Adjustment", International Archives of Photogrammetry and Remote Sensing, Amsterdam, Vol. 33, pp. 461-468, 2000.

Sester, M. and Förstner, W., "Object Location Based on Uncertain Models", Mustererkennung 1989, Informatik Fachberichte 219, Springer-Verlag, pp. 457-464, 1989.

Suveg, I. and Vosselman, G., "3D Reconstruction of Building Models", International Archives of Photogrammetry and Remote Sensing, XXXIII, Part B2, pp. 538-545, 2000.

Tseng, Y.-H., and Wang, S. , "Semiautomated Building Extraction Based on CSG Model-Image Fitting", Photogrammetric Engineering & Remote Sensing 69, 171-180, 2003.

Veldhuis, H., "Performance Analysis of Two Fitting Algorithms for The Measurement of Parameterized Objects", IAPRS Commission 3, Working Group 4, vol. 32, pp. 400-407, 1998.

Voegtle, T., and Steinle, E. , ”3D Modelling of Buildings Using Laser Scanning and Spectral Information”, In International Archives of Photogrammetry and Remote Sensing, vol. XXXIII. pp. 927-934 , 2000.

Vosselman, G., "Interactive Alignment of Parameterised Object Models to Images", IAPRS Commission 3, Working Group 3, vol. 32, pp. 272-278, 1998.

Vosselman, G. and Veldhuis, H., "Mapping by Dragging and Fitting of Wire-Frame Models", Photogrammetric Engineering & Remote Sensing, 65(7), pp. 769-776, 1999.