臺灣博碩士論文加值系統

高解析度影像技巧隨著科技進步而不斷地發展，所對應的硬體規格也隨之提升。使用者需要更精確的影像擷取感測器與更大的儲存記憶體來滿足需求，這代價是十分昂貴的。為了降低取得高解析度影像的成本，利用「超解析度重建技術」將低解析度影像轉換成高解析度影像，或是將較小的影像提升取樣至較大影像的技術不斷地被研發與改進。
在這本篇論文中，我們提出一個基於學習之整合式超解析度影像方法。此方法主要是利用資料庫中高解析度片塊與低解析度片塊相對應的關係所建立出來的模型，去預測估計出低解析度輸入影像中所缺少的細節部分。我們的系統分成兩大部分：訓練階段以及合成階段。在訓練階段，我們會建立一個資料庫；在合成階段，我們會先取得適合的資料並建立自我相似模型然後更新資料庫。接著根據影像片塊的性質去選擇相對應的超解析度演算法，再利用反投影技巧去滿足全域重建限制，最後強化超解析度影像的細節以得到高解析度影像。
相較於現有以學習為基礎的超解析度方法，我們的方法是非常有效率的，而且大大地提高了影像的品質，不僅具有銳利的邊緣以及豐富的細節。

Nowadays, the requirement for image resolution increases fiercely. However, the cost of high resolution images obtained from those modern devices is usually expensive, and it is not easy for people to afford. Therefore, the techniques called “super-resolution” enhancing the low resolution image to higher one are quite important. In recent decades, many researches were dedicated in this field and plenty of algorithms were proposed.
In this thesis, we present an integrated learning-based super-resolution. Learning-based super-resolution techniques model the co-occurrence patterns between the high and low resolution patches of example images to estimate the missing details for low resolution input. Our system has two parts: training phase and synthesis phase. In the training phase, we construct a database. And in synthesis phase, we retrieve some suitable data and build multi-scale self-similarity model to update the database. We choose corresponding super-resolution algorithms based on different content, and we use back-projection to enforce global reconstruction constraint, and then enhance details of the super-resolved image.
Comparing to existing learning-based approaches, our proposed method significantly improves image quality, and the produced super-resolution images have sharp edges and rich details; moreover, the algorithm is very efficient.

口試委員會審定書 #
誌謝 i
中文摘要 ii
ABSTRACT iii
CONTENTS iv
LIST OF FIGURES vi
Chapter 1 Introduction 1
1.1 Introduction of Super-Resolution 1
1.2 Thesis Organization 4
Chapter 2 Related Work 5
2.1 Super-Resolution Method Overview 5
2.2 Learning-Based Super-Resolution 6
2.2.1 Example-Based Super-Resolution 6
2.2.2 Locally Linear Embedding 8
2.2.3 Sparse Representation Method 10
2.2.4 Exploiting Self-Similarities 11
Chapter 3 Integrated Learning-Based Super-Resolution 13
3.1 System Overview 13
3.2 Training Phase 15
3.3 Synthesis Phase 19
3.3.1 Build Multi-Scale Self-Similarity Model 21
3.3.2 Local Reconstruction Constraint 24
3.3.3 Global Reconstruction Constraint 27
3.4 Detail Enhancement 29
Chapter 4 Experimental Results 32
4.1 Classification and statistic 33
4.2 Compared with interpolation-based method and MAP 34
4.3 Compared with other learning-based methods 40
4.4 Compare the processing time 52
4.5 More results 54
Chapter 5 Conclusion and Future Work 59
5.1 Conclusions 59
5.2 Future work 60
REFERENCE 61

[1]S. C. Park, M. K. Park, M. G. Kang, Super-resolution image reconstruction: a technical overview, IEEE Signal Processing Magazine, 2003.
[2]S. Baker, T. Kanade, Limits on super-resolution and how to break them, PAMI, 2002.
[3]M. Irani, S. Peleg, Improving resolution by image registration, CVGIP, 1991.
[4]M. Irani, S. Peleg, Motion analysis for image enhancement: Resolution, occlusion and transparency, JVCI, 1993.
[5]R. R. Schultz, R. L. Stevenson, A bayesian approach to image expansion for improved definition, TIP, 1994.
[6]R. C. Hardie, K. J. Barnard, E. A. Armstrong, Joint map registration and high resolution image estimation using a sequence of undersampled images, TIP, 1997.
[7]M. E. Tipping, C. M. Bishop, Baysian image super-resolution, NIPS, 2003.
[8]J. Sun, N. N. Zheng, H. Tao, Image hallucination with primal sketch priors, CVPR, 2003.
[9]Y. HaCohen, R. Fattal, D. Lischinski, Image upsampling via texture hallucination, ICCP, 2010.
[10]P. Gajjar, M. Joshi, New learning based super-resolution: Use of DWT and IGMRF prior, TIP, 2010.
[11]W. Freeman, T. Jones, E. Pasztor, Example-based super-resolution, CGA, 2002.
[12]H. Chang, D. Y. Yeung, Y. Xiong, Super-resolution through neighbor embedding, CVPR, 2004.
[13]J. Yang, J. Wright, T. Huang, Y. Ma, Image super-resolution via sparse representation of raw image patches, CVPR, 2008.
[14]D. Glasner, S. Bagon, M. Irani, Super-Resolution from a Single Image, CVPR, 2009.
[15]C. Y. Yang, J. B. Huang, M. H. Yang, Exploiting Self-similarities for single frame super-resolution, ACCV, 2010.
[16]W. Freeman, E. Pasztor, O. Carmichael, Learning low-level vision, IJCV, 2000.
[17]J. Wang, S. Zhu, Yihong Gong, Resolution enhancement based on learning the sparse association of image patches, Pattern Recognition Letters, 2010.
[18]S. Lazebnik, C. Schmid, Jean Ponce, Beyond bags of features: Spatial pyramid matching for recognizing natural scene categories, CVPR, 2006.
[19]A. Buades, B. Coll, J. M. Morel, A review of image denoising algorithms, with a new one, SIAM MMS, 2005.
[20]S. Arya, D. M. Mount, Approximate nearest neighbor queries in fixed dimensions, SODA, 1993.
[21]F. Durand, J. Dorsey, Fast bilateral filtering for the display of high-dynamic-range images., SIGGRAPH, 2002.
[22]K. He, J. Sun, X. Tang, Guided Image Filtering, ECCV, 2010.