臺灣博碩士論文加值系統

最近幾年，數位影像抽圖技術已經在影像處理的應用中廣受重視，給定影像C，它可視為前景物件F和背景B，透過方程式C=αF+(1-α)B
組合而成，其中α稱為組合參數，所謂抽圖問題便是如何從影像C中，找出F、B和α。在本論文中，我們首先探討改良的貝氏法，實驗結果顯示，只要trimap事先選取得夠精確，則可得到不錯的抽圖結果，其次，二次規劃法(QP)被用來解決抽圖問題，為了讓求解唯一，我們須再使用QP之前，事先在影像上標示區別前景和背景之筆跡資料，再者，為了降低計算量，給定的影像被切割成多塊子影像，然後在每個子影像上，使用QP求解參數α，最後，許多實驗被執行來證實QP法在影像抽圖問題上的有效性。

In recent years, the digital image matting method has received great attentions in the applications of image processing. Given a digital image C, it can be composed of foreground object F and background B by the matting equation C=αF+(1-α)B, where α is called matting parameter. The matting problem is how to determine F, B and α from C. In this thesis, the modified Bayesian method is first studied. Experimental results reveal that the performance of Bayesian method depends on the choice of the prescribed trimap. Then, the quadratic programming (QP) method is used to solve this matting problem. To let the solution be unique, the users need to specify some scribbles on the image before QP is applied. Moreover, to reduce the computation load of QP method, the given image is partitioned into several sub-images, then each sub-image is processed with QP method to get the desired matting parameter α. Finally, the experiments are conducted to demonstrate the effectiveness of the proposed QP approach in image matting problem.

中文摘要•••••••••••••••••••••••Ⅰ
英文摘要•••••••••••••••••••••••Ⅱ
誌謝•••••••••••••••••••••••••Ⅲ
目錄•••••••••••••••••••••••••Ⅳ
表目錄••••••••••••••••••••••••Ⅴ
圖目錄••••••••••••••••••••••••Ⅵ
第壹章、緒論•••••••••••••••••••••1
1.1研究動機與目的•••••••••••••••••1
1.2相關研究••••••••••••••••••••4
1.3論文架構••••••••••••••••••••10
第貳章、Bayesian抽圖技術•••••••••••••••11
2.1顏色取樣••••••••••••••••••••12
2.2顏色分群••••••••••••••••••••14
2.3求解α值••••••••••••••••••••17
2.4實驗結果••••••••••••••••••••30
第參章、基於二次規劃的影像抽圖方法••••••••••38
3.1二次規劃法介紹•••••••••••••••••38
3.2灰階影像使用二次規劃法抽取物件•••••••••40
3.3實驗結果••••••••••••••••••••64
第肆章、結論與未來展望••••••••••••••••72
4.1結論••••••••••••••••••••••72
4.2未來展望••••••••••••••••••••73
第伍章、參考文獻•••••••••••••••••••74

[1] T. Porter and T. Duff ,“ Compositing digital images”
, In SIGGRAPH, pages253-259, July,1984.

[2]A. R. Smith and J. F. Blinn ,“Blue screen matting”,In Proceedings of SIGGRAPH, pages 259–268, Aug,1996.

[3] Y. Mishima ,“Soft edge chroma-key generation based upon hexoctahedral color space”. U.S. Patent 5,355
,174,1993.

[4] A. Berman, A. Dadourian, and P. Vlahos,“Method for removing from an image the background surrounding a selected object”,U.S. Patent 6,134,346,2000.

[5] M. A. Ruzon and C. Tomasi ,“Alpha estimation in natural images”, In CVPR, pages 18–25, June ,2000.

[6] Y. Chuang , B. Curless , D. Salesin , and R. Szeliski ,“A Bayesian approach to digital matting”, In Proc. CVPR
,2001.

[7]J. Sun , J. Jia , C.-K. Tang , and H.-Y. Shum ,“Poisson matting”, ACM Trans Graph, 23(3):pages315-321
,2004.

[8]A.Levin , D.Lischinski , and Y.Weiss ,“A closed form solution to natural image matting”, in Proc. CVPR, Volume 1, 17-22 ,Page(s):61– 68, June, 2006.
[9] A.Levin , D.Lischinski , and Y.Weiss , “A closed form solution to natural image matting”, Pattern Analysis and Machine Intelligence, IEEE Transactions on Volume 30
,Issue2,Page(s):228-242, Feb.,2008

[10]A.Levin , A.Rav-Acha , and D.Lischinski ,“Spectral matting”, Pattern Analysis and Machine Intelligence, IEEE Transactions on Volume 30, Issue 10, Page(s):1699 – 1712
, Oct. 2008.

[11]Mikhail Sindeyev , Vadim Konushin , and Vladimir Vezhnevets ,“Improvements of Bayesian Matting”,Moscow State Lomonosov University,page(s):23-27,June,2007.

[12]T. Kobayashi , T. Hosaka , N. Otsu,“Image Matting in the Framework of Quantification IV”,in Proc.ICIP , Volume 6, Page(s):VI - 57 - VI – 60, Oct., 2007.

[13] D.Singaraju , R. Vidal,“Interactive image matting for multiple layers”,in Proc.CVPR,23-28, Page(s):1 – 7, June, 2008.

[14] Y. Chuang , New Models and Methods for Matting and Compositing , University of Washington,博士論文，2004。

[15] Jue Wang and Michael F.Cohen, Image and Video Matting : A Survey ,now,2008.

[16] 謝杰融，適應性之物件萃取方法應用於變異光線及陰影環境中的研究，國立東華大學，碩士論文，2005。