本研究旨在提出一計算複雜度低的新型多視角全景圖生成演算法，傳統多視角全景圖生成方法若遇到較有深度變化且距離較近的場景需要密集拍攝以確保品質，使用的影像張數高且資訊使用量低；影像縫合則因視點需固定而無法涵蓋太寬的場景，因此需移動攝影機，但若移動攝影機則會造成相鄰影像的重疊區域中會有嚴重的接合瑕疵。兩方法用於近景拍攝，分別在時間以及品質方面不佳。
　　本論文所提出之方法，首先輸入由一沿著特定直線移動且無旋轉的攝影機所拍攝到的序列影像，每次讀取相鄰兩張影像並分別取得其SURF特徵點並進行配對，接著再將錯誤配對予以去除。利用剩餘的配對，得到兩影像間的平均位移量。在每次讀入的第一張影像中藉由找出上下界決定其裁切量。每張影像都執行相同操作，並將裁剪影像合成一全景圖。
　　本方法在近景拍攝下，所需的影像比傳統多視角全景圖的方法來得低，也因此演算法執行次數較低，速度較快；而本方法產生之全景圖的品質又優於由影像縫合所產生的全景圖。

　　This study aims to propose a novel method of rapid multi-perspective panorama that does not involve complicated calculations such as image transformation and depth evaluation.
　　First, a pair of serial images taken by a straight-moving camera is inputted and SURF feature extraction, feature matching and mismatch removal are executed subsequently. The remaining matches are used to calculate the average displacement between 2 adjacent images. Then the calculation of the Upper Bound and Lower Bound of the first image is executed to decide the part of the image that will be used to form part of the panorama. Repeat the above steps on each image, and the cut images are placed together to generate the desired multi-perspective panorama.
　　Our method requires less required images to form a panorama than the traditional multi-perspective method does and the quality of the generated panorama is better than that from image stitching.

[1]Richard Szeliski, Computer Vision: Algorithms and Applications
[2]Matthew Brown, David G. Lowe., “Automatic Panoramic Image Stitching using Invariant Features,” International Journal of Computer Vision, 2007, 74(1), 59–73.。
[3]Luo Juan, Oubong Gwun., “SURF applied in Panorama Image Stitching,” 2nd IPTA International Conference, 495-499, 2010.
[4]Hyung Il Koo, Nam Ik Cho, "Feature-based image registration algorithm for image stitching applications on mobile devices," Consumer Electronics, IEEE Transactions on, vol.57, no.3, pp.1303-1310, August 2011.
[5]David G. Lowe, "Distinctive image features from scale-invariant keypoints," International Journal of Computer Vision, 60, 2, pp. 91-110, 2004.
[6]M. A. Fishler and R. C. Bolles., “Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography,” Tech report 213, AI Center, SRI International, 1980.
[7]Jiang Yu Zheng, Digital Route Panoramas, IEEE MultiMedia, v.10 n.3, p.57-67, July 2003.
[8]A. Agarwala, M. Agrawala, M. Cohen, D. Salesin, and R. Szeliski., “Photographing long scenes with multi-viewpoint panoramas,” International Conference on Computer Graphics and Interactive Techniques, pages 853–861, 2006.
[9]A. Roman, G. Garg and M. Levoy, “Interactive design of multi-perspective images for visualizing urban landscapes,” IEEE Visualization, pp. 537–544, 2004.
[10]Wei Wang, Hui Gao, Maojun Zhang, Zhihui Xiong, "Multi-perspective Panorama Based on the Improved Pushbroom Model," Digital Media and its Application in Museum & Heritages, Second Workshop on , vol., no., pp.85,90, 10-12 Dec. 2007.
[11]H. Bay, T. Tuytelaars, and L. Van Gool, “SURF: Speeded Up Robust Features,” Proceedings of European Conference on Computer Vision, 2006, pp. 404-417.
[12]P. Viola and M. Jones, “Rapid Object Detection Using a Boosted Cascade of Simple Features,” Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol.2, 2003, pp. 1403-1410.
[13]K. Mikolajczyk and C. Schmid, “A Performance Evaluation of Local Descriptors,” Pattern Analysis and Machine Intelligence, IEEE Transactions on, vol. 27, no. 10, 2005, pp. 1615–1630.
[14]MATLAB R, Available online: http://www.mathworks.com/products/matlab
[15]OpenCV, Available online: http://opencv.org/