簡易檢索 / 詳目顯示

回結果列表
研究生： 許正杰
Cheng-Chieh Hsu
論文名稱： 利用方向性內插法以及運動補償內插法之混合式去交錯系統
Hybrid De-interlacing System using Directional Interpolation and Motion Compensation
指導教授： 陳永昌
Yung-Chang Chen
口試委員:
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： 2004
畢業學年度： 92
語文別： 英文
論文頁數： 77
中文關鍵詞： 去交錯處理 、運動補償 、方向性內插
外文關鍵詞： De-interlacing, Motion compensation, Directional interpolation
相關次數： 點閱：2  下載：0 
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

    • 摘要

    交錯式掃描的技術被使用在傳統的電視系統上以節省傳送資料的頻寬，但是交錯式掃描的信號會產生視覺效果上的不良效應，如邊緣閃爍，線條閃爍，線條效應。去交錯處理是一種能將交錯掃描信號轉換成步進式掃描信號的技術，這個技術已經被廣泛地使用來降低交錯式掃描信號的不良視覺效應，而近年來出產的高畫質電視能支援步進式掃描信號的輸出，因此為了得到較佳的視覺效果，我們需要研發較佳的去交錯處理技術。在這篇論文裡，我們提出了一個以方塊為基礎的去交錯處理演算法。
    提出的以方塊為基礎的去交錯演算法結合了時間運動補償法和空間去交錯補償法，並且提供了能對抗快速運動物體的強健性。對於每個被處理的影像方塊，我們先偵測這個方塊的移動特性，再依照不同的特性做不同方式的去交錯處理，對於快速移動的區塊，我們用大量的空間去交錯處理，對於慢速移動或靜止的區塊，我們用大量的運動補償和時間去交錯處理。對於影像中複雜度較高的部分，我們也使用運動補償和時間去交錯處理。此外，我們的演算法有做攝影機移動的偵測，在攝影機移動的情況之下我們使用運動補償去交錯處理以得到較好的輸出效果。
    根據實驗結果，和傳統的多去交錯處理演算法相比，我們提出的去交錯處理演算法對於大部分影片可以得到比較好的去交錯效果。對於某些測試資料，我們的方法甚至能比傳統的方法提高10dB以上的PSNR。

    Abstract

    Interlaced scanning technique is used for traditional television system to save the bandwidth of the transmitted data. But, uncomfortable visual artifacts such as edge flicker, line crawling, and interline flicker occur due to the inherent nature of the interlaced scanning process. De-interlacing is a technique which can convert the interlaced pictures to progressive pictures. This technique has been widely used to reduce the visual artifact caused by interlaced scanning process. Moreover, recent HDTV systems support the progressive scan to improve the visual quality. Thus, we have to find a good de-interlaced algorithm to get better visual quality. In this thesis, a blocked based de-interlaced algorithm is proposed.
    The proposed block-based de-interlacing method combines motion compensated scheme with spatial de-interlaced scheme and provides better robustness for fast moving object. For each to-be-processed block, we first detect the motion property of this block, and then perform suitable de-interlaced process for each empty pixel in this block. In the case of fast moving block, we use largely spatial de-interlacing. In the case of slow moving or static block, we use largely motion compensated de-interlacing and temporal de-interlacing. For complex regions in an image, motion compensated de-interlacing is performed. Moreover, the proposed algorithm provides camera motion frame detection. For all empty pixels in a camera motion frame, we use motion compensated de-interlacing to get better performance.
    From the simulation result, we see that the proposed de-interlacing algorithm can get better performance than that of traditional algorithms. Our method can even outperform traditional methods about 10dB for some test sequences.

    Table of Contents Abstract i Table of Contents ii List of Figures iv List of Tables vi Chapter 1: Introduction 1 1.1 Overview of interlacing 1 1.2 Motivation 2 1.3 Thesis organization 2 Chapter 2: Overview of De-interlacing Algorithm 4 2.1 The problems and objectives 4 2.2 Conventional de-interlacing systems 8 2.2.1 Non-motion-compensated de-interlacing 8 2.2.2 Motion-compensated de-interlacing 13 2.2.3 Hybrid method 17 2.3 The main challenge 20 Chapter 3: Hybrid Method using Directional Interpolation and Motion Compensation 20 3.1 Algorithm of the proposed method 20 3.1.1 Block-based robust motion detection 20 3.1.2 Two-step motion search with fast motion detection 22 3.1.3 Motion vector smoothness 25 3.1.4 Complex region detection 26 3.1.5 Camera motion detection 27 3.1.6 Directional interpolation using ELA 28 3.1.7 Interpolation method 29 3.1.8 Summary of the proposed algorithm 33 3.2 Architecture design of the proposed algorithm 36 3.2.1 Field buffers and block buffers 37 3.2.2 Field-based processor 40 3.2.3 Block-based de-interlacing module 41 3.2.4 Summary of the proposed architecture 55 Chapter 4: Simulation Result 56 4.1 Simulation environments 56 4.1.1 Measurement system 56 4.1.2 De-interlacing parameters 57 4.2 Simulation result 57 4.2.1 Foreman sequence 58 4.2.2 Container sequence 60 4.2.3 Mother and daughter sequence 61 4.2.4 Stefan sequence 63 4.2.5 Flower garden sequence 65 4.2.6 Coastguard sequence 67 4.2.7 Pendulum sequence 69 4.3 Summary 73 Chapter 5: Conclusions and Future Works 74 5.1 Conclusions 74 5.2 Future works 75 References 76 List of Figures Fig. 1-1 Interlaced video 1 Fig. 1-2 The process of de-interlacing 3 Fig. 2.1-1 The sampling lattices of progressive scanning and interlaced scanning 4 Fig. 2.1-2 The carrier lattices of progressive scanning and interlaced scanning 5 Fig. 2.1-3 The replications of the continuous spectrum after progressive sampling and interlaced sampling 6 Fig. 2.1-4 The spectrums of video signals after sampling 6 Fig. 2.2-1 The required frequency domain passbands of VT-filtering 9 Fig. 2.2-2 VT filtering de-interlacing technique 9 Fig. 2.2-3 Motion adaptive de-interlacing algorithm 10 Fig. 2.2-4 An example for block-based motion detection 10 Fig. 2.2-5 A local window for the ELA interpolation 11 Fig. 2.2-6 De-interlacing using motion compensated field insertion 13 Fig. 2.2-7 De-interlacing using motion compensated average 14 Fig. 2.2-7 Switching of motion vector 15 Fig. 2.2-9 De-interlacing using hybrid method proposed in [11] 17 Fig. 2.2-9 De-interlacing using MCFI proposed in [3] 18 Fig. 3.1-1 (a) The conventional block-based motion detection method 21 Fig. 3.1-1 (b) Misjudgement results from fast motion object 21 Fig. 3.1-2 The proposed robust motion detection 21 Fig. 3.1-3 An example of video sequence with fast moving object 22 Fig. 3.1-4 Two-step motion with fast motion detection 23 Fig. 3.1-5 Bi-directional refined search 24 Fig. 3.1-6 Neighboring blocks used for MV smoothing 25 Fig. 3.1-7 Complex region detection 26 Fig. 3.1-8 Camera motion detection 28 Fig. 3.1-9 The seven-point ELA algorithm 29 Fig. 3.1-10 Fast motion information detection 30 Fig. 3.1-11 Input pixels of median filter 30 Fig. 3.1-12 Deteermination of “BadMotion” and “Edge” 32 Fig. 3.1-13 Flow chart of choosing final interpolation method 32 Fig. 3.1-14 Flow chart of proposed algorithm (a) 34 Fig. 3.1-15 Flow chart of proposed algorithm (b) 35 Fig. 3.2-1 Block diagram of the proposed architecture 36 Fig. 3.2-2 The architecture of a field-buffer 37 Fig. 3.2-3 Connection between frame buffers and field buffers 38 Fig. 3.2-4 Block buffer 39 Fig. 3.2-5 Flow chart of “MBcounter” controller 40 Fig. 3.2-6 Block diagram of “block-based de-interlacing” module 41 Fig. 3.2-7(a)(b) IO ports and block diagrams of “Motion Detection module” 41 Fig. 3.2-7(c) Architecture of “Motion Detection processor” 43 Fig. 3.2-8 IO ports of Initial motion search module and Refined search module 43 Fig. 3.2-9 Block diagram of “Initial motion search” module 44 Fig. 3.2-10 Architecture of “MVI processor” 45 Fig. 3.2-11 Block diagram of “Refined motion search” module 46 Fig. 3.2-12 Architecture of “MVR processor” 47 Fig. 3.2-13 IO ports of “MV smoothing” module 48 Fig. 3.2-14 IO ports of “MV smoothing” module 49 Fig. 3.2-15 Architecture of “MVS processor” 49 Fig. 3.2-16 IO ports of camera motion detection 50 Fig. 3.2-17 Block diagram of camera motion detection module 51 Fig. 3.2-18 Architecture of “CPX Processor” module 51 Fig. 3.2-19 Functions of “J” and “counter” 52 Fig. 3.2-20 Block diagram of the “ELA & Median interpolation” module 52 Fig. 3.2-21 The block diagram of “Processor_N” module 53 Fig. 3.2-22 The block diagram of “ELA” module 54 Fig. 3.2-23 The block diagram of “OutPixel” module 54 Fig. 4-1 Performance measurement by re-converting 56 Fig. 4.2-1 The PSNR performance of the four methods on the “Forman” sequence 58 Fig. 4.2-2 154th reconstructed image of test sequence “Foreman” 59 Fig. 4.2-3 The PSNR performance of the four methods on the “container” sequence 60 Fig. 4.2-4 264th reconstructed frame of test sequence “Container” 61 Fig. 4.2-5 The PSNR performance of the four methods on the “Mother and daughter” sequence 62 Fig. 4.2-6 58th reconstructed image of test sequence “Mother and daughter” 61 Fig. 4.2-7 The PSNR performance of the four methods on the “Stefan” sequence 63 Fig. 4.2-8 176th reconstructed image of test sequence “Stefan” 64 Fig. 4.2-9 The PSNR performance of the four methods on the “Flower garden” sequence 65 Fig. 4.2-10 13th reconstructed image of test sequence “Flower Garden” 66 Fig. 4.2-11 The PSNR performance of the four methods on the “Coastguard” sequence 67 Fig. 4.2-12 220th reconstructed image of test sequence “Coastguard” 68 Fig. 4.2-13: The 4th (a) and 5th (b) interlaced images of the “Pendulum” sequence 69 Fig. 4.2-14: Reconstructed images of the “Pendulum” sequence using different methods 70 Fig. 4.2-15: Reconstructed images of the “Toilet paper” sequence. 71 Fig. 4.2-16: Reconstructed images of the “Flag” sequence. 72 List of Tables Table. I The parameters of the proposed algorithm 57 Table. II PSNR comparison 73

    [1] G. de Haan and E.B.Bellers, “De-interlacing-an overview”, Proceedings of the IEEE, vol86, Issue 9, pp. 1839-1857, Sept.1998.
    [2] G. de Haan and E.B.Bellers, “De-interlacing of Video Data”, IEEE Transactions on Consumer Electronics, Vol.44 Issue 3, pp.819-825, Aug. 1997.
    [3] You-Young Jung, Byung-Tae Choi, Yung-Jun Park, and Sung-Jea Ko, “An Efficient De-interlacing Technique Using Motion Compensation Interpolation”, IEEE Transations on Consumer Electronics, Vol 46 Issue 3, pp 460-466,Aug. 2000.
    [4] Byung-Tae Choi, Sung-Hee Lee, and Sung-Jea Ko, “New Frame Rate Up-conversion using Bi-directional Motion Estimation”, IEEE Transactions on Consumer Electronics, Vol.46, No.3, August 2000.
    [5] G. de Haan and E.B.Bellers, “De-interlacing A Key Technology for Scan Rate Conversion”, ELSEVIER, 2000.
    [6] Alexander E. Mohr, Eve A. Riskin and Richard E. Ladner, “Generalized Multiple Description Coding through Unequal Loss Protection,” Proceedings of IEEE International Conference on Image Processing, vol.1, pp.411-415, Dec. 1999.
    [7] Thomas Komarek and Peter Pirsch, “Array Architecture for Block Matching Algorithms”, IEEE Transactions on Circuits and Systems, Vol. 36, No. 10, October 1989.
    [8] Kenju Sugiyama and Hiroya Nakamura, “A Method of Deinterlacing with Motion Compensation Interpolation”, IEEE Transactions on Consumer Electronics, Vol.45, No.3, August 1999.
    [9] Yeong-Taeg Kim, Shin-Haeng Kim, and Se-Woomg Park, “Motion Decision Feedback Deinterlacing Algorithms”, IEEE International Conference on Image Processing, 2002..
    [10] Hoon Yoo and Jechang Jeong, “Direction-Oriented Interpolation and Its Application to De-interlacing”, IEEE Transactions on Consumer Electronics, Vol.48, no. 4, November 2002.
    [11] Ohjae Kwon, Kwanghoon Sohn, and Chulhee Lee, “Deinterlacing using Directional Interpolation and Motion Compensation”, IEEE Transactions on Consumer Electronics, Vol.49, No. 1, February 2003.
    [12] You-Young Jung, Seungjoon Yang, and Pillho Yu,“An Effective De-Interlacing Technique Using two Types of Motion Information”, IEEE Transactions on Consumer Electronics, Vol.49, No.3, August 2003.
    [13] Stephan Wenger, “Error Patterns for Internet Experiments,” ITU Telecommunications Standardization Sector, Red Bank, New Jersey, Doc. Q15-I-16r1, Oct. 1999.
    [14] Shyh-Feng Lin, Yu-Ling Chang, and Liang-Gee Chen, “Motion Adaptive Interpolation with Horizontal Motion Detection for Deinterlacing”, IEEE Transactions on Consumer Electronics, Vol.49, No.4, November 2003.
    [15] Rob A. Beuker and Imran A. Shah, “Analysis of Interlaced Video Signals and Its Applications”, IEEE Transactions on Image Processing, Vol. 3, No. 5, Septemper 1994.
    [16] Renxiang Li, Bing Zeng, and Ming L. Liou, “Reliable motion Detection/Compensation for Interlaced Sequences and Its Applications to Deinterlacing”, IEEE Transactions on Circuits and Systems for Video Technilogy, Vol. 10, No. 1, February 2000.

    無法下載圖示 全文公開日期 本全文未授權公開 (校內網路)
    全文公開日期 本全文未授權公開 (校外網路)

    QR CODE