Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/19690
標題: 使用梯度學習之強健單張超解析度影像演算法
Robust Single Image Super Resolution Method Using Gradient-Based Learning
作者: 王仁助
Wang, Jen-Chu
關鍵字: super-resolution
超解析度
machine learning
gradient operators
Sobel filter
機器學習
梯度運算子
索貝爾濾波器
出版社: 資訊科學與工程學系所
引用: [1]S. C. Park, M. K. Park and M. G. Kang, “Super-Resolution Image Reconstruction: A Technical Overview,” IEEE Signal Processing Magazine, vol. 20, NO.3 (May 2003), pp. 21-36. [2]W. T. Freeman, T. R. Johes and E. C. Pasztor, “Example-Based Super-Resolution,” IEEE Computer Graphics and Application, vol. 22, NO. 2 (March 2002), pp.56-65. [3]R. Y. Tsai and T. S. Huang, “Multiframe image restoration and registration,” Advances in Computer Vision and Image Processing, vol. 1, JAI Press, London, 1984,pp. 317-339. [4]H. Strak and P. Oskoui, “High resolution image recovery from image-plane arrays, using convex projection,” Journal of Optical Society of America A, vol. 6, 1989, pp. 1715-1726 [5]M. Irani and S. Peleg, “Improving resolution by image registration,” CVGIP: Graphical Models and Image Processing, vol. 53, pp. 231-239, May 1991. [6]R. R. Schulz and R. L. Stevenson, “Extraction of high-resolution frames from video sequences,” IEEE Transaction on Image Processing, vol. 5, NO. 6, pp.996-1001, June 1996. [7]W. T. Freeman, E. C. Pasztor and O. T. Carmichael, “learning low-level vision,” International Journal of Computer Vision, vol. 40, No. 1, pp. 25-47, October 2002. [8]D. Datsenko and M. Elad, “Example-based single document image super-resolution: a global MAP approach with outlier rejection,” Multidimensional Systems and Signal Processing, vol. 18, pp. 103-121, September 2007. [9]T. M. Chan, J. Zhang, J. P. and H. Huang, “Neighbor embedding based super-resolution algorithm through edge detection and feature selection,” Pattern Recognition Letters, vol. 30, pp.494-502, 2009. [10]C. V. Jiji, M. V. Joshi and S. Chaudhuri, “Single-Frame Image Super-Resolution Using Learned Wavelet Coefficients,” International Journal of Imaging Systems and Technology, vol. 14, no. 3, pp.105-112, April 2004. [11]C. V. Jiji and S. Chaudhuri, “Single-frame image super-resolution through contourlet learning,” EURASIP Journal on Applied Signal Processing, vol. 2006, pp. 1-11,2006. [12]X. Li, K. M. Lam, G. Qiu, L. Shen and S. Wang, “Example-based image super-resolution with class-specific predictors,” Journal of Visual Communication and Image Representation, vol. 20, pp. 312-322,2009. [13]Y. Linde, A. Buzo and R. M. Gray, “An Algorithm for Vector Quantizer Design,” IEEE Transactions on Communications, vol. 28, No.1, pp.84-95, January 1980. [14]G. Qiu, “A progressively predictive image pyramid for efficient lossless for coding,” IEEE Transactions on Image Processing,” vol. 8, No.1, pp.109-115, 1999.
摘要: 超解析度(super-resolution)影像就是利用單張或是多張的低解析度影像,透過學習或是重建的方法產生出一張具有高品質且富有豐富細節資訊的高解析度影像技術。高解析度的影像在很多應用上都是極度被需求的,包括數位相機、衛星影像、醫學影像、以及犯罪調查等。舉例來說,具有高解析度的醫學影像能提高醫生診斷的準確率以及手術的成功率,而在犯罪調查方面,高解析度的影像則能有效提供警方更多的辦案線索。 多張的低解析度影像超解析度影像重建演算法,經過影像校正、影像內插和影像復原的技術就可以重建一張高解析影像。對於採用多張的低解析度影像超解析度影像重建演算法都有相同的前提假設,必須對一個場景拍攝多張的低解析度影像,但是在很多的情況下我們並不容易對相同的場景擷取多張的影像,所以後來有學者提出採用單張的低解析度影像超解析度影像重建演算法,然而單張的低解析度影像假若直接採用影像內插技術,我們只能得到一張缺少了細節的高解析度影像,所以必須額外有一個擁有多張的高解析度影像資料庫,透過搜尋或是學習的方法找到被放大後的低解析度影像所缺少的細節,最後將細節加入被放大後的影像中,就可以得到擁有豐富細節資訊的高解析度影像。 大多數單張的低解析度影像超解析度影像演算法,都是採用搜尋的方式,利用低解析度影像中不同的特徵做為搜尋的條件,從影像資料庫內尋找最相近或是匹配的低解析度影像細節,再將其所對應的高解析度細節補回被放大後的影像當中,就可以得到高解析度影像。但是採用收尋的方式會因為影像資料庫的多元性和大小而對執行效率與結果有直接的影響,當影像資料擁有更多的資訊,相對收尋時間也同樣是等比例的成長,所以Li採用機器學習的方式,於訓練的階段透過學習得到低解析度影像細節與高解析度影像細節的關連性,測試的階段只需將低解析度的影像細節輸入已訓練完成的預測器,就可以更快速與精確的得到所對應的高解析度影像細節。 但是Li的方法容易受到雜訊干擾而使得到的高解析度影像受到嚴重的破壞,而大部分的數位影像在取得或是傳送過程會因為感光元件或是傳送頻帶而使得影像受到雜訊的干擾,所以我們改良Li所提出來的超解析度影像演算法,利用梯度學習來解決因為雜訊影響而使得高解析度影像受到嚴重破壞的問題。
Super-resolution is a technique that produced a high-resolution image enlarged from low-resolution image with more detail. Example-based approach using the corresponding patch pairs between high-resolution and low-resolution difference image of example images to estimate the missing detail for single low-resolution input. In this paper, we present Gradient-Based learning method learned the relations between the high-resolution difference patches and low-resolution gradient patches for super-resolution to improve the weakness result of similar learning method suffer from damage slightly noise influence. Experimental results show the effectiveness and robustness of the proposed method.
URI: http://hdl.handle.net/11455/19690
其他識別: U0005-1607201014124800
文章連結: http://www.airitilibrary.com/Publication/alDetailedMesh1?DocID=U0005-1607201014124800
Appears in Collections:資訊科學與工程學系所

文件中的檔案:

取得全文請前往華藝線上圖書館



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.