超解析度近來在影像處理領域已經是一個流行的研究主題，所謂的超解析度是從單張或多張低解析度的影像放大得到高解析度影像的一個過程，由於近來電子設備的演進，超解析度則是一個很重要的議題。
在超解析度領域中，一個很著名的方法是利用自我相似以及影像金字塔的方式，也就是利用不同的影像大小中找出相似的影像區塊，利用這些影像區塊去合成較高解析度的影像。
然 而 這 方 法 因 為 相 似 區 塊 的 不 足 跟 每 層 放 大 導 致 的 誤 差 (Error Propagation)，導致在某些物件上產生不真實的形狀以及邊緣，為了克服這些問題，我們在這邊提出一個超解析度方法的架構，利用原有影像金字塔架構，額外用影像插補法產生另外一個金字塔架構，並且比較彼此區塊的相似度。我們希望藉由減少每一層的誤差，來讓更高層的影像結果更好。
最後我們的實驗結果證明我們的方法可以達到更高的峰值訊噪比，除此之外也可以有效果的改善視覺品質。

Super-resolution has been a popular research topic in the image processing area.
It is a process of getting a high-resolution image from one or multiple low-resolution
images. We focus on the super resolution method that produces a high resolution
image from a single low resolution image, which is method with self-similarity image
pyramid. In the algorithm, we can find the similar patch in the image pyramid to patch
up the correspond patch to obtain a high resolution image.
However, the method could result in unrealistic shape and edge on the certain
object in the reconstructed image due to the insufficient patch and the error
propagation. To overcome the problem, we use the framework of self-similarity image
pyramid and compared with weighted interpolation method in each layer, and improve
upper layers’ accuracy by reducing lower layers’ error.
Our experimental results show that proposed method could reach higher PSNR
value than some existing methods and improve visual quality in the reconstructed
image.

[1] R. Keys, "Cubic convolution interpolation for digital image processing," IEEE
TranS. Acoust., Speech, Signal Process., vol. 29, no. 6, pp. 1153-1160, Dec. 1981.
[2] X. Li and M. T. Orchard, “New edge-directed interpolation,” IEEE
Trans. Image Process., vol. 10, no. 10, pp. 1521–1527, Oct. 2001.
[3] W. T. Freeman, T. R. Jones and E. C. Pasztor, "Example-Based
Super-Resolution," IEEE Computer Graphics and Applications, vol. 22, no. 2, pp.
56-65, 2002.
[4] Sam T. Roweis, Lawrence K. Saul, “Nonlinear Dimensionality Reduction
by Locally Linear Embedding”, Science 2000, vol. 290 no. 5500 pp. 2323-2326,
Dec 2000.
[5] H. Chang, D.Y. Yeung and Y. Xiong, “Super-resolution through neighbor
embedding”. CVPR, 2004, pp. I-275 - I-282.
[6] J. Yang, J. Wright, T. Huang, and Y. Ma, “Image super-resolution as sparse
representation”, IEEE Transactions on Image Processing, vol.19, no.11, pp.
2861-2873.Nov. 2010.
[7] D. Glasner, S. Bagon and M. Irani, "Super-resolution from a single image,"
in ICCV, 2009, pp. 349-356
[8] G. Freedman and R. Fattal, "Image and video upscaling from local self-
examples," ACM Trans. Graph., vol. 30, no. 2, p. 12, April 2011.
[9] C. Yang, J. Huang , M. Yang ,”Exploiting Self-Similarities for Single Frame
Super-Resolution” ACCV2011, pp. 497-510
[10] Y. Chen, Y. Gao and Liu, K.J.R, “An evolutionary game-theoretic approach for
image interpolation”.IEEE Conf. ICASSP, 2011,pp. 989-992.
[11] R. Fisher, The genetical theory of natural selection, Clarendon Press,
Oxford, 1930.
[12] S. Arya and D. M. Mount. “Approximate nearest neighbor queries in fixed
dimensions”. In SODA, 1993. pp. 271-280
[13] M. Irani and S. Peleg. “Improving resolution by image registration”.
CVGIP,1991. pp. 231-239., April 1991