簡易檢索 / 詳目顯示

回結果列表
研究生： 徐芳凱
Hsu, Fang-Kai
論文名稱： 快速紋理導向搜尋及高效率預測之樣本基礎超解析度演算法
Fast Texture Oriented Search and Effective Prediction in Example-based Super Resolution
指導教授： 謝明得
Shieh, Ming-Der
學位類別： 碩士
Master
系所名稱： 電機資訊學院 - 電機工程學系
Department of Electrical Engineering
論文出版年： 2016
畢業學年度： 104
語文別： 英文
論文頁數： 59
中文關鍵詞： 單影像超解析度演算法 、樣本基礎超解析度演算法 、局部二值模式 、局部多層次梯度模式 、加強預測 、旋轉不變性
外文關鍵詞： Single-image super-resolution, Example-based super-resolution, Local binary pattern, Local multi-gradient level pattern, Enhanced prediction, Rotation invariance
相關次數： 點閱：92  下載：1 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

    • 單影像超解析度演算法（SISR）被廣泛應用於高解析度顯示器的相關產品上，在可能的解決方案中樣本基礎的方法雖可以預先訓練資料來提供各種影像細節，透過回歸學習的方式，儲存的樣本可以被訓練成轉換函數進而大量減少儲存空間，但如何快速搜索最合適的轉換函數仍是此系統的瓶頸。本論文提出一個利用局部多層次梯度模式來有效地描述樣本上的局部幾何，並且通過一個簡單的查詢表將樣本分類到相對應的群組。
    影像重建的結果可以透過結合各種預測來更進一步地改善，本論文針對使用加強預測提出一個高效率的重建方案，當使用所提出的局部多層次梯度模式來做樣本分群，其加強預測的重建過程就可以預先完成。此應用除了可以在不增加其時間複雜度的條件下進一步改善影像重建的品質，亦能用來大幅縮減轉換函數的儲存空間。最後，本論文也使用了軟性決定的方法來補償分群造成的錯誤。
    本論文使用峰值信噪比(PSNR)和保真度準則(IFC)來比較所提出的超解析度演算法和其他七個先進演算法的優劣。從實驗結果顯示，所提出的演算法使用了最少的執行時間，若和七個演算法中最快的相比，只需其84%的執行時間就能得到差不多的表現，若再加上軟性決定的機制，整體的實驗結果更勝於這個主要的競爭對手。

    Single-image super-resolution (SISR) has been adopted for high-resolution display related applications. Example-based approaches can provide rich image details by using trained dataset. Of the existing example-based solutions, regression-based methods can be applied to reduce the required memory storage size by training mapping functions instead of using a huge dictionary. However, the speed of searching the nearest cluster for the desired mapping function remains one of the bottlenecks of the system. This thesis first presents an operator denoted as local multi-gradient level pattern to fast yet effectively describe the patch local geometry. The corresponding cluster can then be quickly identified by a simple lookup table.
    The reconstructed image quality can be further enhanced by combining various prediction results. This thesis also proposes an effective model reconstruction method for enhanced prediction. The desired model can be constructed offline when using the presented local multi-gradient level pattern as the clustering feature. Applying the proposed schemes can further improve the quality of reconstructed high-resolution image while retaining almost the same time complexity as the original solution. Moreover, the space for storing mapping functions can be dramatically reduced by using the proposed model combining method. Finally, this thesis also proposes a soft decision scheme to alleviate the side effect of accessing an improper cluster due to unexpected conditions such as noisy images.
    The proposed SR method is intensively compared with seven state-of-the-art SR methods in terms of the peak signal-to-noise ratio (PSNR) and information fidelity criterion (IFC). Experimental results show that the proposed one demands the shortest run time and takes only 84% execution time of the fastest SR solution in the seven competitors with close performance. The image quality can be further improved by employing the proposed soft decision scheme.

    摘 要..........................i ABSTRACT ....................... iii 致謝.........................v Content…. ....................... vi List of Tables .................... viii List of Figures ..................... ix Chapter 1. Introduction .................. 1 1.1 Motivation ..................... 1 1.2 Related works .................... 2 1.3 Thesis organization .................. 4 Chapter 2. Background .................... 5 2.1 General structure of example-based super-resolution ........ 5 2.2 LR-to-HR mapping function in dictionary type ........... 8 2.2.1 Clustered example-based super-resolution ........... 9 2.2.2 Neighbor embedding and sparse representation .......... 11 2.3 Direct mapping approach ................. 14 2.3.1 Anchored neighbor regression .............. 14 2.3.2 Patch clustering and matrix function ............ 16 2.4 Rotation invariance in image ................ 19 Chapter 3. Proposed Fast Texture Oriented Search .......... 21 3.1 System overview ................... 21 3.1.1 Feature extraction and luminance normalization ........ 23 3.1.2 The prepared matrix table ............... 24 3.2 Proposed texture oriented clustering scheme ........... 25 3.2.1 Local binary patterns ................. 25 3.2.2 Two-stage local binary patterns ............. 26 3.2.3 Local multi-gradient level pattern ............ 27 3.3 Effective model construction for enhanced prediction ........ 28 3.3.1 Rotation invariance characteristic in LBP ........... 28 3.3.2 Simplification and enhanced prediction ............ 29 3.3.3 Reduction for memory .................. 32 3.4 Soft decision for quantization .............. 35 3.4.1 Concept of soft decision ................ 35 3.4.2 Constraints for applying soft decision ............ 37 3.4.3 Weights of matrix combining ............... 38 Chapter 4. Experimental Results and Comparison .......... 39 4.1 Analysis for the proposed patch clustering schemes ........ 40 4.1.1 Setting suitable interval value for 2SLBP ........... 41 4.1.2 Parameter setting for LMGP ................ 42 4.1.3 Rarely occurring clusters ............... 44 4.2 Improvement by combining predictions in proposed clustering ..... 46 4.2.1 Comparison of enhanced prediction and data augmentation ..... 46 4.2.2 The optimized combined weight for soft decision ......... 48 4.3 Performance comparison .................. 50 Chapter 5. Conclusion and Future Work ............. 55 5.1 Conclusion ...................... 55 5.2 Future work ..................... 56 References….. ..................... 57

    [1] W. Freeman, T. Jones and E. Pasztor, "Example-based super-resolution", IEEE Comput. Grap. Appl., vol. 22, no. 2, pp. 56-65, 2002.
    [2] J. Choi and M. Kim, "Super-Interpolation With Edge-Orientation-Based Mapping Kernels for Low Complex 2× Upscaling", IEEE Transactions on Image Processing, vol. 25, no. 1, pp. 469-483, 2016.
    [3] T. Chan, J. Zhang, J. Pu and H. Huang, "Neighbor embedding based super-resolution algorithm through edge detection and feature selection", Pattern Recognition Letters, vol. 30, no. 5, pp. 494-502, 2009.
    [4] F. Yeganli, M. Nazzal and H. Ozkaramanli, "Image super-resolution via sparse representation over multiple learned dictionaries based on edge sharpness and gradient phase angle", Signal, Image and Video Processing, vol. 9, no. 1, pp. 285-293, 2015.
    [5] H. Yin, S. Li, and J. Hu, "Single image super resolution via texture constrained sparse representation," in ICIP, 2011, pp. 1161-1164.
    [6] H. Chang, D.-Y. Yeung, and Y. Xiong. "Super-resolution through neighbor embedding" CVPR, 01:275–282, 2004.
    [7] M. Bevilacqua, A. Roumy, C. Guillemot, and M.-L. Alberi Morel. "Low-complexity single-image super-resolution based on nonnegative neighbor embedding" In BMVC,2012.
    [8] J. Yang, J. Wright, T. Huang, and Y. Ma. "Image super-resolution via sparse representation" Image Processing, IEEE Transactions on, 19(11):2861–2873, 2010.
    [9] R. Zeyde, M. Elad, and M. Protter. "On single image scale-up using sparse representations" In Curves and Surfaces, pages 711–730, 2010.
    [10] M. Aharon, M. Elad, and A. Bruckstein. "K-SVD: an algorithm for designing overcomplete dictionaries for sparse representation." IEEE Trans. on Sig. Proc., 54(11), 2006.
    [11] J. A. Tropp and A. C. Gilbert. "Signal recovery from random measurements via orthogonal matching pursuit" IEEE Transactions on Information Theory, 53(12), 2007.
    [12] R. Timofte, V. De and L. V. Gool, "Anchored neighborhood regression for fast example-based super-resolution", Proc. IEEE Int. Conf. Comput. Vis., pp. 1920-1927, 2013.
    [13] R. Timofte, V. De Smet and L. Van Gool, "A+: Adjusted anchored neighborhood regression for fast super-resolution,"Proc. IEEE Asian Conf. Comput. Vis., pp. 111-126, 2014.
    [14] C.- Y. Yang and M.-H. Yang, "Fast direct super-resolution by simple functions", ICCV, 2013.
    [15] Jeong, Shin-Cheol, and Byung Cheol Song. "Training-based super-resolution algorithm using k-means clustering and detail enhancement" Signal Processing Conference, 2010 18th European. IEEE, 2010.
    [16] D. Dai, R. Timofte and L. Van Gool, "Jointly optimized regressors for image superresolution" Eurographics, vol. 7, pp. 8, 2015.
    [17] R. Timofte, R. Rothe and L. Van Gool, "Seven ways to improve example-based single image super resolution", arXiv:1511.02228, 2015.
    [18] T. Ojala, M. Pietikäinen, and T. Mäenpää, "Multiresolution gray-scale and rotation invariant texture classification with local binary patterns" IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, no. 7, pp. 971-987, 2002.
    [19] H.R. Sheikh, A.C. Bovik, and G. de Veciana, "An information fidelity criterion for image quality assessment using natural scene statistics", IEEE Trans. on Image Processing, vol. 14, no. 12, pp. 2117-2128, 2005.
    [20] F. Yeganli, M. Nazzal, M. Unal and H. Ozkaramanli, "Image super-resolution via sparse representation over multiple learned dictionaries based on edge sharpness", Signal, Image and Video Processing, vol. 10, no. 3, pp. 535-542, 2015.
    [21] S. Yang, M. Wang, Y. Chen and Y. Sun, "Single-Image Super-Resolution Reconstruction via Learned Geometric Dictionaries and Clustered Sparse Coding", IEEE Transactions on Image Processing, vol. 21, no. 9, pp. 4016-4028, 2012.
    [22] M. Cimpoi, S. Maji, I. Kokkinos, S. Mohamed, and A. Vedaldi. "Describing textures in the wild" In Proc. CVPR, 2014.
    [23] C. W. Chen, F. K. Hsu, D. W. Yang, J. Wang, and M. D. Shieh, "Fast model searching and combining for example learning-based super-resolution" in Proc. IEEE Int. Symp. Circuits Syst. May. 2016.
    [24] R. B. Marimont and M. B. Shapiro, "Nearest neighbor searches and the curse of dimensionality" Journal of the Institute of Mathematics and Its Applications, 24:59–70, 1979.
    [25] A. N. Tikhonov and V. Y. Arsenin. Solution of Ill-posed Problems. Winston & Sons, 1977.
    [26] N. Dalal and B. Triggs "Histograms of oriented gradients for human detection" Coference on Computer Vision and Pattern Recognition (CVPR), 2005.
    [27] X. Li and M. T. Orchard "New edge-directed interpolation" IEEE Trans. Image Process., vol. 10, no. 10, pp. 1521–1527, Oct. 2001.
    [28] C. Dong, C. C. Loy, K. He, and X. Tang "Learning a deep convolutional network for image super-resolution" in Proc. 13th Eur. Conf. Comput. Vis., Zurich, Switzerland, Sep. 2014, pp. 184–199.
    [29] Cui, Zhen, et al "Deep network cascade for image super-resolution" European Conference on Computer Vision. Springer International Publishing, 2014.
    [30] Kim, Jiwon, Jung Kwon Lee, and Kyoung Mu Lee "Accurate image super-resolution using very deep convolutional networks" arXiv preprint arXiv:1511.04587 (2015).
    [31] Wang, Zhaowen, et al. "Deep networks for image super-resolution with sparse prior." Proceedings of the IEEE International Conference on Computer Vision. 2015.
    [32] Kim, Jiwon, Jung Kwon Lee, and Kyoung Mu Lee. "Deeply-Recursive Convolutional Network for Image Super-Resolution." arXiv preprint arXiv:1511.04491 (2015).
    [33] Pérez-Pellitero, Eduardo, et al. "Antipodally Invariant Metrics for Fast Regression- Based Super-Resolution

    下載圖示 校內：2021-07-21公開
    校外：2021-07-21公開
    QR CODE