立體視覺技術(Stereo Vision)可以結合生活中許多應用，像前幾年一度很流行的3D電影，還有最近一些行車偵測技術也會結合立體視覺技術。立體視覺技術主要是由立體匹配產生出對應的深度資訊，在演算法方面又分為區域性和全域性，一般而言區域性演算法可以達到速度快、計算量小的優點，而全域性演算法優點為可以產生出更準確的深度資訊。為了達到即時性的需求，本文選擇了區域性演算法，可是要克服深度資訊不準確的缺點，我們使用了平均濾波器(Mean Filter)和引導影像濾波器(Guided Image Filter)來優化深度資訊。一般而言，區域性演算法分為四個階段：匹配代價計算(Matching Cost Computation)、匹配代價聚合(Cost Aggregation)、視差選擇(Disparity Selection)、視差值最佳化(Disparity Refinement)，引導影像濾波器可以應用於匹配代價聚合或是視差值最佳化的階段，本文探討在這兩個階段應用的相互關係和影響。在平均濾波器的硬體設計中，我們使用了Moving-Sum的方式來取代積分影像，影像讀取除了使用一般的掃描方式(Line-based)，也設計了另一種讀取法(Stripe-based)的硬體版本，以節省更多的記憶體使用量。

Stereo vision has many applications, including 3D movies and the recent advanced driver assisted systems (ADAS). In stereo vision, stereo matching of generating depth information is the most critical technique. In general, there are two categories of stereo matching methods: global and local. Local stereo matching methods are fast due to less computation while global methods can generate more accurate depth information at the cost of more computation complexity. This thesis uses local stereo matching methods with mean filtering and guided image filtering to improve the quality of depth information. A generic local stereo matching method can be divided into four stages: matching cost computation, cost aggregation, disparity selection, and disparity refinement. Guided image filtering can be applied to cost aggregation or disparity refinement. This thesis will study the impacts of using guided image filtering in different stereo matching stages. In hardware implementation for involved mean filtering, instead of the conventional integral image method, we use the moving-sum method with different data reading schemes of with line-based or stripe-based and compare the hardware resource requirement of internal memory buffers

摘要 i
Abstract ii
目錄 iii
圖目錄 v
表目錄 vii
第1章、 導論 1
1.1 研究動機 1
1.2 論文架構 1
第2章、 研究背景與相關研究 3
2.1 立體匹配背景 3
2.1.1 立體視覺成因 3
2.1.2 極線幾何與極線幾何限制（Epipolar Geometry） 5
2.2 立體匹配演算法 7
2.2.1 匹配代價計算 7
2.2.2 匹配代價聚合 8
2.2.3 視差選擇 9
2.2.4 視差值最佳化 12
第3章、 演算法流程 13
3.1 匹配代價計算 14
3.2 匹配代價聚合 16
3.3 視差選擇 16
3.4 視差值最佳化 17
3.4.1 左右一致性檢查 17
3.4.2 引導影像濾波器 18
3.4.3 加權中值濾波 22
第4章、 硬體架構設計 25
4.1 整體架構 25
4.2 匹配代價計算 26
4.2.1 色彩影像轉亮度影像 26
4.2.2 梯度計算 27
4.2.3 Cost Volume Construction 28
4.3 匹配代價聚合 33
4.3.1 平均濾波器之硬體設計 33
4.3.2 Moving sum 與積分影像之比較 35
4.4 視差選擇 37
4.5 視差值最佳化 38
4.5.1 左右一致性檢查之硬體設計 38
4.5.2 引導影像濾波器之硬體設計 40
4.5.3 加權中值濾波 43
4.6 硬體設計第二版 46
4.6.1 匹配代價計算 49
4.6.2 左右一致性硬體第二版 51
第5章、 實驗結果與數據比較 53
5.1 效能評比 53
5.2.1 匹配代價聚合硬體選擇之影響 53
5.2.2 引導影像濾波器硬體優化之影響 54
5.2.3 除法器和移位器之影響 54
5.2 邏輯合成數據與分析 55
5.2.1 FPGA數據分析 55
5.2.2 Design Compiler 數據分析 57
5.3 論文比較 57
5.4 測試影像 59
第6章、 結論與未來展望 63
6.1 結論 63
6.2 未來展望 63
參考文獻 64

[1] D. Scharstein and R. Szeliski, “A taxonomy and evaluation of dense two-frame
stereo correspondence algorithms,” Int. J. Comput. Vision,vol. 47, pp. 7–42, 2002
[2]N. Y.-C. Chang, et al., “Algorithm and Architecture of Disparity Estimation With Mini-Census Adaptive Support Weight,＂in IEEE Transactions on Circuits and Systems for Video Technology,vol. 20, no. 6, June. 2010
[3] D. Scharstein, “View Synthesis Using Stereo Vision”, Dissertation of Cornell
University PHD, 1997
[4] W.-L. Wang, “Hardware Design for Disparity Estimation Using Dynamic Programming,” Master''s thesis, University of NSYSU, 2012
[5] J-M, Huang ,“Hardware Design of Disparity Estimation Using Belief Propagation,” Master''s thesis, University of NSYSU, 2013
[6] A. Hosni, et al., “Fast cost-volume filtering for visual correspondence
and beyond,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 35,no. 2, pp. 504–511, Feb. 2013.
[7] A.Hosni, et al., “Real-time local stereo matching using guided image filtering” IEEE International Conference on Multimedia and Expo, pp. 1 – 6, 2011
[8] K. He, J. Sun, and X. Tang, “Guided Image Filtering,” IEEE Trans.
Pattern Anal. Mach. Intell., vol. 35, no. 6, pp. 1397–1409, Jun. 2013.
[9] C. Ttofis; C. Kyrkou and T. Theocharides,“A Low-Cost Real-Time Embedded Stereo Vision System for Accurate Disparity Estimation Based on Guided Image Filtering”, IEEE Transactions on Computers, Vol. 65, 2016,
[10] Z. Ma, et al., “Constant time weighted median filtering for stereo matching and beyond,” in Proc. Int. Conf. Comput. Vis, pp. 49–56. , 2013.
[11] Qi Zhang; Li Xu and Jiaya Jia “100+ Times Faster Weighted Median Filter (WMF) ”, IEEE Conference on Computer Vision and Pattern Recognition., pp. 2830-2837, 2014
[12]C. Ttofis; C. Kyrkou and T. Theocharides, “High-Quality Real-Time Hardware Stereo Matching Based on Guided Image Filtering.” Design, Automation & Test in Europe Conference & Exhibition (DATE), pp. 1-6., 2014
[13]Pauline Tan, Pascal Monasse. “Stereo Disparity through Cost Aggregation with Guided Filter.Image Processing On Line”, IPOL - Image Processing on Line, pp.252-275., 2014
[14]Sichao Wang and Tsutomu Maruyama , “An Implementation method of the Box Filter on FPGA. ” 2016 26th International Conference on Field Programmable Logic and Applications (FPL), pp.1-8., 2016
[15]Xuchong Zhang, et al., “VLSI Architecture Exploration of Guided Image Filtering for 1080P@60Hz Video Processing. ” IEEE Transactions on Circuits and Systems for Video Technology, pp. 1-1., 2017
[16] Chieh-Chi Kao; Jui-Hsin Lai and Shao-Yi Chien “VLSI Architecture Design of Guided Filter for 30 Frames/s Full-HD Video.” IEEE Transactions on Circuits and Systems for Video Technology, pp. 513 - 524., 2014
[17] Chen Yang; Yan Li; Wei Zhong and Song Chen “Real-Time Hardware Stereo Matching Using Guided Image Filter.” 2016 International Great Lakes Symposium on VLSI (GLSVLSI) , pp. 105 - 108., 2016
[18] Simon Perreault and Patrick Hebert “Median Filtering in Constant Time” IEEE Transactions on Image Processing, vol. 16, pp. 2389-2394., 2007
[19] Lazaros Nalpantidis and Antonios Gasteratos “Review of Stereo Vision Algorithms:From Software to Hardware.” International Journal of Optomechatronics, vol. 2, pp. 435-462., 2008
[20] S. A. Fahmy; P. Y. K. Cheung and W. Luk “Novel FPGA-based Implementation of Median and Weighted Median filters for Image Processing” International Conference on Field Programmable Logic and Applications, 2005, pp. 142-147, 2005
[21] W. Wang, et al., “Real-Time High-Quality Stereo Vision System in FPGA. ” IEEE Transactions on Circuits and Systems for Video Technology, vol. 25 , pp. 1696–1708,2015
[22] Huahua Chen, “Stereo Matching Using Dynamic Programming Based on Occlusion Detection. ” 2007 International Conference on Mechatronics and Automation, pp. 2445–2449,2007