無論是MPEG或H.26X系列皆為現今主要的動態影像壓縮標準，但不管為哪一種標準，動態補償都為其核心原理之一。在本論文中，我們提出透過一種非方塊的動態補償來與傳統方塊式的動態補償做比較，並觀察其畫面效果好壞及資料量多寡。
傳統標準MPEG裡，是對整張圖進行掃描，並以固定16*16大小的方塊為單位進行動態補償的計算，進而判斷方塊是否產生位移、紀錄差值，在特殊的情況下，真實物體形狀用方塊來切割並不適合，且資料重複和龐大。在實驗中分為兩階段，第一階段為影像區域的分割，將動態物體的區域切割出來，第二階段得變形還原所需的對應參數，用以記錄最小資料量。
我們在現有的MPEG-1軟體系統中實現將動態補償改用optical flow(OF, 光流)和幾何轉換來代替，再用形態運算來做修補，並在解碼端成功還原播放用以觀察結果。發現最後還原效果不理想，資料壓縮效果還也未達到期望。

MEPG and H.26X series are two major standards of video compression. Motion compensation is an important method in the above standards. In this thesis, we follow a new method of non-block motion compensation proposed by our Lab before to compare with traditional block motion compensation. We study the visual quality and data compression for both methods.
In the traditional MPEG, we scan whole image to calculate the motion compensation with fixed 16*16 block as an unit. Furthermore, we record the difference the shift of blocks. In some specific circumstances, it is not suitable to separate the real shape of object by blocks, which may cause a huge data as a result. There are two stage in our study. In stage one, we use optical flow (OF) to separate the area of motion object. In stage two, we get the corresponding parameters for geometry transformation to record a small amount of data.
We do our experiments by replacing motion compensation in the original MPEG-1 system. We also modify the images by mathematical morphology. It was found that the final restoration effect and the data compression effect was not as satisfactory as expected.

論文審定書 i
致謝 ii
摘要 iii
Abstract iv
目錄 v
圖表目錄 vii
第一章 導論 1
1.1 研究動機 1
1.2 研究方式 1
1.3 章節介紹 3
第二章 MPEG介紹 4
2.1 MPEG簡介 4
2.2 MPEG-1壓縮原理 5
第三章 理論背景 11
3.1 簡介 11
3.2 光流之基本原理 11
3.2.1 光流的定義 11
3.2.2 理論基礎 14
3.2.3 計算方式 17
3.3 區域分割 21
3.3.1 區域成長 21
3.3.2 區域門檻 22
3.4 幾何轉換 24
3.4.1幾何轉換簡介 24
3.4.2 空間轉換 24
3.4.3 決定灰階 26
3.5 形態運算 27
3.5.2 形態運算的定義 27
3.5.3 擴張與侵蝕運算 27
第四章 實驗背景 29
4.1 簡介 29
4.2產生輪廓區域 32
4.2.1光流切割[9][14] 32
4.2.2幾何轉換變形[9][14] 35
4.3決定幾何參數所需的座標 37
第五章 實驗結果與分析 39
5.1 簡介 39
5.2 輪廓變化和形態運算修補結果比較 40
5.3 傳統動態補償系統和幾何轉換系統之還原結果比較 45
5.4 壓縮率之計算與討論 48
第六章 結論 50
Reference 51
Acronyms 54

[1] Z. Diab and P. Cohen, “Motion Compensated Video Compression Using Adaptive Transformations,” IEEE International Conference on Acoustics, Speech, and Signal Processing(1997 ICASSP), Munich, Germany, pp. 2881-2884, 21-24 April 1997.
[2] K. Li and Y. Yang, “Fire Detection Algorithm Based on CLG-TV Optical Flow Model,” IEEE 2nd International Conference on Computer and Communications (2016 ICCC), Chengdu, China, pp. 1381-1385, 14-17 Oct. 2016.
[3] A. Guerre, M. Lamard, P. H. Conze, B. Cochener, and G. Quellec , “Optical Flow Estimation in Ocular Endoscopy Videos Using Flownet on Simulated Endoscopy Data,” IEEE 15th International Symposium on Biomedical Imaging (2018 ISBI), Washington, DC, USA, pp. 1463-1466, 4-7 April 2018.
[4] J. Cheng, Y. H. Tsai, S. Wang, and M. H. Yang, “SegFlow: Joint Learning for Video Object Segmentation and Optical Flow,” IEEE International Conference on Computer Vision (2017 ICCV) , Venice, Italy, pp. 686-695, 22-29 Oct. 2017.
[5] Y. Yemez, B. Sankur, and E. Anarim, “Region Growing Motion Segmentation and Estimation in Object-oriented Video Coding,” IEEE 3rd International Conference on Image Processing(1996 ICIP), Lausanne, Switzerland, pp. 521-524, 19-19 Sept. 1996.
[6] R. Gonzalez and R. Woods, “Digital Image Processing,” Addison-Wesley Inc, pp. 518-548, 1992.
[7] A. K. Jain, “Fundamentals of Digital Image Processing,” Englewood Cliffs, N.J., Prentice-Hall Inc, 1989.
[8] R. Gonzalez and R. Woods, “Digital Image Processing,” Addison-Wesley Inc, pp. 296-302, 1992.
[9] 周本生, “視訊編碼與標準簡介,”上課講義: 幾何轉換其分支與實現。
[10] J.J. Gibson, “On the Analysis of Change in the Optic Array,” Scandinavian Journal of Psychology(1977 SJoP), Wiley Online Library, pp. 161-163, 1977.
[11] K. Nakayama and J.M. Loomis, “Optical Velocity Patterns, Velocity-sensitive Neurons, and Space Perception:A Hypothesis,” Perception 3, pp. 63-80, 1974.
[12] B. G . Schunck, “Image Flow : Fundamentals and Algorithm,” Motion Understanding: Robot and Human Vision, W.N. Martin and J. K. Aggarwal, Eds. Norwell, Kluwer Academic Publishers, 1988.
[13] D. Sun, S. Roth, and M. J. Black, “Secrets of Optical Flow Estimation and Their Principles,” IEEE Computer Society Conference on Computer Vision and Pattern Recognition(2010 CVPR), San Francisco, CA, USA, pp. 2432-2439, 13-18 June 2010.
[14] 周本生, “視訊編碼與標準簡介,” 上課講義: 光流,分割其分支與實現。
[15] S. Rinsurongkawong, M. Ekpanyapong and, M. N. Dailey, “Fire Fetection for Early Fire Alarm Based on Optical Flow Video Processing,” 9th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (2012 ECTI-CON), Phetchaburi, Thailand, pp. 1–4, 16-18 May 2012.
[16] 周本生, “視訊編碼與標準簡介,” 上課講義:檔案之管理資料與內容資料其分支與實現。
[17] Z. Liu and J. Luo, “An optical flow method for elastography at large compression using three image frames,” IEEE International Ultrasonics Symposium (2017 IUS), Washington, DC, USA, pp. 1-4, 6-9 Sept. 2017.