(3.236.222.124) 您好!臺灣時間:2021/05/13 02:07
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果

詳目顯示:::

: 
twitterline
研究生:何家誌
研究生(外文):Jia-Jhih HE
論文名稱:以動態向量資訊運用於單視角視訊之 深度圖產生演算法
論文名稱(外文):Based on The Motion Vector Information for Single-view Depth Map Generation Algorithm
指導教授:李佩君李佩君引用關係
指導教授(外文):Pei-Jun Lee
口試委員:楊家輝郭景明林國祥
口試委員(外文):Jar-Ferr YangJing-Ming GuoGuo-Shiang Lin
口試日期:2013-01-29
學位類別:碩士
校院名稱:國立暨南國際大學
系所名稱:電機工程學系
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2013
畢業學年度:101
語文別:中文
論文頁數:64
中文關鍵詞:2D to 3DMonocularSeed Growing深度圖動態向量
外文關鍵詞:2D to 3DMonocularSeed Growingdepth mapmotion vector
相關次數:
  • 被引用被引用:0
  • 點閱點閱:230
  • 評分評分:系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔
  • 下載下載:6
  • 收藏至我的研究室書目清單書目收藏:1
近年來,人們為了追求立體效果許多3D顯示技術不斷推陳出新。然而,目前市面上的視訊內容還是以2D影像為主流,原因在於3D影像的製作成本昂貴及現存影像依舊以2D影像居多。為達到3D影像之普遍化,除了提昇3D影像製作品質及下降製作成本外,2D影像轉為3D影像系統是目前十分重要的技術。而在轉換技術中,深度圖的品質影響最後的3D效果,因此如何產生出高品質的深度圖將是3D轉換技術中的關鍵。
目前的深度圖產生方法可分為利用單視角影像產生深度圖及雙視角影像產生深度圖本論文進行單視角影像之深度圖產生的方法研究。因為人眼對於動態的物體有較高的興趣,為了在動態物件中提供較佳的3D效果,本論文利用動態向量的資訊來產生深度圖。然而動態向量預估的不準確性將會導致深度圖品質的下降,為了解決此問題,提出以Seed Growing演算法進行物件分割並找尋各物件相對應的特徵點,搭配可調式遮罩的方法增加比對的準確性,另外對前後兩張畫面中相同物件給予相同深度值來解決深度值不連續,用以減少連續畫面中深度值的不連續而造成觀賞者的不舒適性。
在實驗結果的部分,本論文提出之單視角深度圖產生方法在準確率上,與其他方法比較,PSNR的表現上升了0.06~2.93dB,SSIM的評估上升0.01~0.18。另外本論文在主觀性評比上採用了MOS評分,依此觀察,本論文所提出的方法對於畫面當中動量較大的物件有更好的3D效果。

In recent years, researcher develops many 3D technologies for stereo display. However, the content of video is based on 2D images in the market because the manufacturing cost of 3D images is expensive and 2D images exist more than 3D images. To develop 3D technology, the 2D video to 3D video conversion system is the very important technology. Since the quality of the depth map will influences the final 3D effect, the high quality depth map generation is a key scheme in the 2D to 3D conversion system.
Now, the depth map generation methods can be classified into Monocular and Binocular depth generation algorithm. In this thesis, the depth map generation method for Monocular video is studied to improve the quality of depth map. The viewers are more interested in the objects which are in motion. In order to get better 3D effect, the information of motion vectors is used to generate the depth map in this thesis. However, the inaccuracy of motion vectors estimation will lead to declining in the quality of the depth map. For solving the problem, this thesis uses Seed Growing for object segmentation and finding corresponding feature points in each object, and it can be combined with an adaptive matching window which is produced by the cross-based matching algorithm to enhance the matching accuracy. To solve the depth value discontinuing and to reduce the reviewer uncomfortable discontinuing depth value in the continuous frames on the proposed algorithm utilizes the information of the object size to track the same object between the current frame and the previous frame, and gives the same depth value to the same object.
In the experimental result, it is proved that the result with proposed method is better by comparing the performance, it raises about 0.06 to 2.93dB in PSNR, it raises about 0.01 to 0.18 dB in SSIM. In MOS, the object of the high motion vector can get the better 3D effect in the frame by the proposed algorithm.

目錄
誌謝.......................................................................................................................i
論文摘要................................................................................................................................ii
Abstract.................................................................................................................................iv
第一章、緒論........................................................................................................................1
1.1 簡介.........................................................................................................................1
1.2 研究背景.................................................................................................................2
1.3 研究動機.................................................................................................................4
1.4 論文章節組成.........................................................................................................5
第二章、文獻探討與回顧......................................................................................................6
2.1 綜論深度圖產生演算法.........................................................................................6
2.2 單張深度圖產生.....................................................................................................8
2.2.1 基於消失點深度圖產生..............................................................................9
2.2.2 基於邊緣資訊深度圖產生........................................................................12
2.2.3 基於動態向量深度圖產生........................................................................14
2.3 運用雙張深度圖產生法中之比對演算法於單視角視訊深度預估...............18
2.4 結論.......................................................................................................................20
第三章、提出之基於動態向量深度圖產生演算法............................................................21
3.1 以亮度直方圖偵測移動區域...............................................................................22
3.2 物件分割...............................................................................................................24
3.2.1 型態濾波器................................................................................................24
3.2.2 Seed Growing..............................................................................................26
3.3 特徵點檢測...........................................................................................................30
3.4 可調式遮罩大小檢測...........................................................................................31
3.5 動態向量轉換深度值...........................................................................................33
3.6 Depth Value Estimation for Video Sequence.........................................................37
3.7演算法流程與架構...............................................................................................39
第四章、實驗結果與探討....................................................................................................41
4.1 客觀分析...............................................................................................................43
4.2 主觀分析...............................................................................................................57
第五章、結論及未來展望..................................................................................................61
5.1 結論.......................................................................................................................61
5.2 未來展望...............................................................................................................61
參考文獻..............................................................................................................................62











圖目錄
圖 1.1 3D顯示器(a)3D電視[1] (b)3D投影機[2]................................................................1
圖 1.2 立體成像原理...........................................................................................................2
圖 1.3 多視角畫面攝影機[3]..............................................................................................3
圖 2.1 2D to 3D流程圖[9]....................................................................................................6
圖 2.2 彩色影像與其對應的深度圖[10]............................................................................6
圖 2.3 深度圖(a)顏色資訊[8] (b)紋理清晰度[11] (c)線性透視[6]....................................8
圖 2.4 Vanishing Point之分塊流程圖[6]............................................................................9
圖 2.5 消失點示意圖[6]....................................................................................................10
圖 2.6 (a)彩色影像[12] (b)物件分類[12] (c)深度圖[12]...................................................11
圖 2.7 文獻[13]之深度圖產生流程圖..............................................................................13
圖 2.8 (a)video sequence (b)recorder depth map (c)estimated depth map (d)estimated depth map by our approach[14]................................................................................15
圖 2.9 藉由顏色分割後的資訊強化深度圖[15]..............................................................16
圖 2.10 Depth map assignment of foreground[17]..............................................................17
圖 2.11 十字形可調式遮罩示意圖[19]............................................................................19
圖 2.12 不同像素與其對應的可調式遮罩大小[19]........................................................19
圖 3.1 比較目前畫面中的MB與參考畫面中相同位置MB的亮度值...........................22
圖 3.2 對應到圖3.1中兩張圖左邊框框內的亮度直方圖...............................................22
圖 3.3 對應到圖3.1中兩張圖右邊框框內的亮度直方圖...............................................23
圖 3.4 Mobile移動區域檢測(a)原始圖 (b)經過亮度直方圖篩選過後的示意圖...........23
圖 3.5 Kendo移動區域檢測(a)原始圖 (b)經過亮度直方圖篩選過後的示意圖............24
圖 3.6 Kendo影像序列(a)dilation後影像 (b)erosion後影像 (c)物件邊緣之梯度影像..............................................................................................................................25
圖 3.7 Seed Growing...........................................................................................................26
圖 3.8 使用顏色資訊將小區塊標籤合併.........................................................................28
圖 3.9 物件分割(a)原始彩圖影像 (b)Seed Growing分割 (c)利用顏色資訊合併後的影像..............................................................................................................................29
圖 3.10 (a)特徵點檢測 (b)以區塊為單位求得之特徵點.................................................31
圖 3.11 (a)可調式遮罩以特徵點為中心以垂直方向長成 (b)經由垂直方向長成後以N
為中心向左右方向成長........................................................................................32
圖 3.12可調式遮罩示意圖................................................................................................32
圖 3.13 全區域搜尋法搭配可調式遮罩[22]....................................................................34
圖 3.14 攝影機運動放大/縮小識別區域..........................................................................35
圖 3.15 Kendo sequence背景區域動態向量示意圖(a)原始動態向量 (b)透過camera
motion校正之後動態向量....................................................................................36
圖 3.16 Video Depth Estimation流程圖.............................................................................38
圖 3.17以動態向量產生深度圖示意圖 (a)物件分割 (a)動態向量檢測 (c)深度圖產
生............................................................................................................................40
圖 4.1 五組測試影像(a)Kendo (b)Ballet (c)Lovebird1 (d)Mobile (e)Big Buck
Bunny[23]...............................................................................................................42
圖 4.2 利用各種演算法求得之Kendo深度圖(a)原始彩圖 (b)原始深度圖 (c)Proposed algorithm (d)MV of BMA[15] (e)MV of H.264[14]...............................................48
圖 4.3 利用各種演算法求得之Lovebird1深度圖(a)原始彩圖 (b)原始深度圖 (c)Proposed algorithm (d)MV of BMA[15] (e)MV of H.264[14]...........................49
圖 4.4 利用各種演算法求得之Ballet深度(a)原始彩圖 (b)原始深度圖 (c)Proposed algorithm (d)MV of BMA[15] (e)MV of H.264[14]...............................................50
圖 4.5 利用各種演算法求得之Mobile深度圖(a)原始彩圖 (b)原始深度圖 (c)Proposed algorithm (d)MV of BMA[15] (e)MV of H.264[14]...............................................51
圖 4.6 利用各種演算法求得之Big Buck Bunny[23]深度圖(a)原始彩圖 (b)原始深度圖 (c)Proposed algorithm (d)MV of BMA[15] (e)MV of H.264[14]...........................52
圖 4.7 各種影像序列亮度直方圖統計(a)(c)(e)(g)(i)本論文產生之深度 (b)(d)(f)(h)(j)亮度直方圖統計......................................................................................................54
圖 4.8 Kendo物件分割圖(a)Seed Growing (b)分水嶺演算法[25] ..................................55
圖 4.9 Lovebird1物件分割圖(a)Seed Growing (b)分水嶺演算法[25] ............................55
圖 4.10 Ballet物件分割圖(a)Seed Growing (b)分水嶺演算法[25] .................................55
圖 4.11 Mobile物件分割圖(a)Seed Growing (b)分水嶺演算法[25] ...............................56
圖 4.12 Big Buck Bunny[23]物件分割圖 (a)Seed Growing (b)分水嶺演算法[25] ........56
圖 4.13 立體影像測試流程[26] .......................................................................................57
圖 4.14 測驗立體影像的評分標準[26] ...........................................................................57
圖 4.15 各影像序列之立體感直方圖 (a)Kendo (b)Lovebird1 (c)Ballet (d)Mobile (e)Big Buck Bunny[23].......................................................................................................59












表目錄
表 4.1 測試影像特性說明.................................................................................................41
表 4.2 Test Sequence PSNR(dB)分析.................................................................................44
表 4.3 Test Sequence SSIM分析........................................................................................45
表 4.4 Test Sequence 深度圖產生之時間統計(秒/Frame) ..............................................45
表 4.5主要動作執行之時間統計(秒/Frame)....................................................................46
表 4.6 Big Buck Bunny[23]PSNR(dB)及SSIM數據分析.....................................46
表 4.7 深度圖之立體感MOS平均分數表.......................................................................58

[1] http://www.tpbb.com.tw/apply.asp?lv=1&id=268
[2] http://www.viewsonic.com.tw/products/projectors/Pro8300.php
[3] http://graphics.stanford.edu/projects/array/
[4] C. Fehn, “A 3D-TV system based on video plus depth information,” in Preceedings of
Conference Record of the Thirty-Seventh Asilomar Conference on Signals, System and
Computers, volume 2, pp.1529-1533, 2003.
[5] P. kauff, N. Atzpadin, C.Fehn, M. Muller, O. Schreer, A. smolic and R. Tanger,
“Depth Map Creation And Image-based Rendering for Advanced 3DTV Services
Providing Interoperability and Scalability,” Image Communications, vol.22, no.2, pp.217-234, Feb. 2007.
[6] Yi-Min Tsai, Yu-Lin Chang and Liang-Gee Chen, “Block-based Vanishing Line
and Vanishing Point Detection for 3D Scene Reconstruction,” in Proceeding of
International Symposium on Intelligent Signal Processing and Communications,
pp. 586 – 589, Dec. 2006.
[7] Donghyun Kim, Dongbo Min, Kwanghoon Sohn,”A Stereoscopic Video
Generation Method Using Stereoscopic Display Characterization and Motion
Analysis,” IEEE Transactions on Broadcasting, vol. 54, pp. 188-197, June. 2008.
[8] Sheng-Po Tseng and Shang-Hong Lai, “Accurate Depth Map Estimation from Video
via MRF Optimization,” 2011 IEEE Visual Communications and Image Processing
(VCIP), PP.1-4, Nov. 2011.
[9] Yeong-Kang Lai, Yu-Fan Lai and Ying-Chang Chen, “An Effective Hybrid
Depth-Generation Algorithm for 2D-to-3D Conversion in 3D Displays,” Journal of
Display Technology, volume: PP, pp.1-8, 2012
[10] Liang Zhang, Carlos Vázquez and Sebastian Knorr, “3D-TV Content Creation: Automatic 2D-to-3D Video Conversion,” IEEE Transactions on Broadcasting, vol. 57, pp.372-383, June 2011.
[11] Kamining He, Jian Sun, and Xiaoou Tang, “Single Image Haze Removal Using
Dark Channel Prior,” IEEE Transactions on Pattern Analysis and Machine
Intelligence, vol.33, pp.2341-2353, 2011
[12] J. I. Jung, and Y. S. Ho, “Depth Map Estimation from Single-view Image using
Object Classification Based on Bayesian Learning,” 3DTV-Conference: The True
Vision - Capture, Transmission and Display of 3D Video (3DTV-CON), 2010,
pp.1-4, 7-9 June 2010.
[13] 鄭朝鐘 (陳良基 教授指導) , “三維影像訊號處理之演算法和架構設計”,國
立台灣大學電子所碩士論文,2010年10月
[14] M. T. Pourazed, P. Nasiopoulos, and R. K. Ward, “An H.264-based Scheme for 2D to
3D Video Conversion,” ICCE’09. Digest of Technical Papers International
Conference on Consumer Electronics, 2009, vol.55, pp.742-748, May. 2009.
[15] Lai-Man Po, Xuyuan Xu, Yuesheng Zhu, Shihang Zhang, Kwok-Wai Cheung
and Chi-Wang Ting, ”Automatic 2D-to-3D video conversion technique based on
depth-from-motion and color segmentation,” 2010 IEEE 10th International
Conference on Signal Processing(ICSP), pp. 1000-1003, Oct. 2010.
[16] Rui Han, and Aidong Men, “MOTION ESTIMATION COMPENSATION FOR
FRAME RATE UP CONVERSION AND 2D-TO-3D CONVERSION AT 1080P
RESOLUSIONS,” Beijing University of Posts and Telecommunications.
[17] Yeong-Kang Lai, Yu-Fan Lai, and Ying-Chang Chen, “An Effective Hybrid
Depth-Generation Algorithm for 2D-to-3D Conversion in 3D Displays,” Journal of
Display Technology, vol.9, pp.154-161, 2013.
[18] Xiaojun Huang, Lianghao Wang, Junjun Huang, Dongxiao Lai, and Ming Zhang, “A
Depth Extraction Method Based On Motion and Geometry for 2D to 3D Conversion,” Third International Symposium on Intelligent Information Technology Application, vol.3, pp.294-298, Nov. 2009.
[19] Ke Zhang, Jiangbo Lu and Lafruit, G., “Cross-Based Local Stereo Matching
Using Orthogonal Integral Images,” IEEE Transactions on Circuits and Systems
for Video Technology, vol. 19, pp.1073-1079, Jul. 2009.
[20] J. Xiao, L. Xia and L. Lin, “ Segment-based stereo matching using edge dynamic
programming,” in Proceedings of International Congress on Image and Signal Processing (CISP), pp. 1676-1679, Wuhan, China, Oct. 2010.
[21] 賴宥成(李佩君 教授指導),“Rate Control Algorithm for Multi-view Video
Coding Using Visual Perception Based Stereo Video Analysis,”國立暨南國際大學
電機所碩士計畫2011年7月.
[22] 莊啟峰(李佩君 教授指導),“Depth Map Generation Algorithm for 2-View to
Multi View Conversion,”國立暨南國際大學電機所碩士計畫2011年7月.
[23] http://www.bigbuckbunny.org/index.php/download/
[24] Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli , “Image quality
assessment: From error visibility to structural similarity” IEEE Transaction on
Image Processing, vol.13, no. 6, pp.600-612, 2004.
[25] 游文傑(李佩君 教授指導),“Depth Map Generation based on the Watershed
Algorithm and its Hardware Implementation,”國立暨南國際大學電機所碩士計畫
2012年7月.
[26] Call for Proposals on 3D Video Coding Technology ISO/IEC JTC1/SC29/WG11,
MPEG2011/N12036, Geneva, Switzerland, Mar. 2011.

QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top
系統版面圖檔 系統版面圖檔