近年來，在數位影像處理領域中，如何從有限的二維影像中還原出三維影像資訊的主題變得越來越重要。一般來說，解決影像還原問題的方法是要用目前已有的條件去反算原本的影像。因為即使是來源完全不同的影像，經過干擾之後其差距也有可能相去不遠，所以這是屬於一種病態問題。而有關影像還原之反算的古典解法是最小平方法，然而這種方法因為需要去運算龐大條件數的線性方程組，所以經常需要耗費相當可觀的計算時間。
本論文提出了一種減少還原運算量的演算法，此演算法利用影像中活動度的差異來預測深度資訊，進一步降低演算法整體所需要的時間。然而這種預測方式可能會造成計算時的失誤。我們在這裡引入去雜訊的觀念來彌補這種失誤，使用西格瑪濾波器跟中位數濾波器，幫我們修正不合理的運算結果。
在論文的最後會將我門的演算法跟改良前的演算法做比較，經實驗結果證明我們的演算法在降低計算量上有相當明顯的效果，而在計算出來的深度圖上也有不輸給原演算法的表現。

In the recent years, the topic of how to recovery 3-Dimension image in digital image processing field is more and more important in digital image processing field. The
solution of the image recovery problem is to calculate the original image by inverse the existed condition. After interrupt, the images from different source may almost the
same, so it is ill-posed problem. The classic solution of the image recovery is leastsquares. However, this approach need to calculate ill-conditioned linear equations, and
it always cost huge computation time.
In this thesis, we propose a algorithm to reduce the computations. For shortening the computation time, the algorithm predicts the result of the depth map by using the distinct from activity regions. However, the rediction may occur miss when calculations. We bring the idea of denoise to restore the failure of the predictions, and sigma filter and mean filter are used to repair the inaccurate result of calculations. Finally, some experiments comparing the proposed method with the original method are presented. Experimental results show that the proposed method has outstanding efficiency on reducing the computations, and the performance of the proposed method is similar to the original method.

[1] Radu Ciprian Bilcu and Markku Vehvilainen. A new image de-noising technique based on image decomposition and sigma filtering. WSEAS Transactions on Communications,
(3):579--586, August 2005.

[2] Paolo Favaro and Stefano Soatto. A geometric approach to shape from defocus.IEEE Transactions on Pattern analysis and Machine Intelligence, 27(3), March 2005.

[3] Akira Kubota and Kiyoharu Aizawa. Reconstructing arbitrarily focused images from two differently focused images using linear filters. IEEE Transactions on
Image Processing, 14(11), November 2005.

[4] Martin Burger Paolo Favaro, Stefano Soatto and Stanley J. Osher. Shape from defocus via diffusion. IEEE TRANSACTION OF PATTERN RECOGNITION AND MACHINE INTELLIGENCE, 30(3), March 2008.

[5] A.N. Rajagopalan and S. Chaudhuri. An mrf model-based approach to simultaneous recovery of depth and restoration from defocused images. IEEE Transactions on Pattern Analysis and Machine Intelligence, 21(7), July 1999.

[6] A.N. Rajagopalan, S. Chaudhuri, and Uma Mudenagudi. Depth estimation and image restoration using defocused stereo pairs. IEEE Transactions on Pattern Analysis and Machine Intelligence, 26(11), November 2004.