博碩士論文 110523063 詳細資訊




以作者查詢圖書館館藏 以作者查詢臺灣博碩士 以作者查詢全國書目 勘誤回報 、線上人數:13 、訪客IP:18.225.98.100
姓名 劉君桓(Chun-Huan Liu)  查詢紙本館藏   畢業系所 通訊工程學系
論文名稱 CNN為架構的分散式影像編碼用於HEVC畫面間預測
(CNN-Based DVC Architecture for HEVC Inter Prediction)
相關論文
★ 10Gb/s MM XFP光收發模組設計與實現★ 資訊產品自動化測試之研究
★ 高電流密度鰭式氮化鎵高電子遷移率電晶體研究★ 電子郵件及壓縮檔案解碼之研究
★ 渦輪碼在光學記錄系統上之應用★ 離散餘弦轉換硬體架構之研究
★ 動態影像之錯誤隱藏研究★ 即時性無失真壓縮編碼之研究
★ 類神經網路在手寫數字辨識之研究★ 事後機率演算法則在資料儲存系統之研究
★ 紅外線傳輸協定及通道之研究★ 低密度同位元檢查碼在數位資料儲存系統之研究
★ 一種新型的JPEG2000竄改偵測與還原技術★ 即時性無失真壓縮之研究
★ 混合快速模式決策演算法之研究★ 光學記錄MEPR2通道系統之時序恢復探討與研究
檔案 [Endnote RIS 格式]    [Bibtex 格式]    [相關文章]   [文章引用]   [完整記錄]   [館藏目錄]   至系統瀏覽論文 (2027-1-22以後開放)
摘要(中) HEVC高效率視頻編碼的編碼結構中,相較於之前的影像壓縮標準,將CTU的大小從最大的64x64切割至8x8的尺寸,降低了編碼單元的位元率,但也增加了計算的時間成本。因此,本研究在第三章提出以卷積神經網路 CNN為主的分散式影像編碼架構應用於HEVC的編碼端和解碼端,來簡化編碼過程的複雜度並在解碼端後處理時提高影像品質。
在編碼端,我們對SVM-CNN CU/PU的演算法進行優化,將原本的差值濾波器替換為雙線性插值濾波器,以減少計算量也節省編碼時間。然而簡化小數點估算導致畫面失真。因此我們使用CNN對畫面進行後處理改善,使BDBR%降低到0.43%,TS%增加到74.42%。而在解碼端,我們加入DenseNet/DAE三通道CNN模型對解碼影像進行畫面增強,使BDBR%能降到-5.96%。
在第四章中,我們探討了如何調整編碼端演算法中的閥值,在影像品質改善的限制下最佳化省時率。透過對SAD和RDO等判別特徵的閥值將其依比例進行調整,得到ΦSAD和ΦRDO分別與BDBR%和TS%的空間分佈。為了更精確預測,我們在BDBR% =6.0%附近進行了更多的實驗。得到的ΦSAD和ΦRDO之間的關係式並帶入ZBDBR和ZTS的曲面函數,計算出我們預測的最佳省時率。最後,我們根據BDBR%和TS%的關係曲線進行預測。結果顯示,預測值和實驗結果的誤差在可接受的範圍內。因此未來我們可以透過調整閥值來優化編碼端的計算,進一步預測出編碼端的BDBR%和TS%效能表現。
摘要(英) In the coding structure of HEVC, compared to previous image compression standards, the size of Coding Tree Units (CTU) has been reduced from a maximum of 64x64 to 8x8, lowering the bit rate of encoding units but increasing the computational time cost. Therefore, in this study, a Distributed Video Coding architecture based on CNN (Convolutional Neural Networks), is proposed for both the encoder and decoder of HEVC. The goal is to simplify the complexity of encoding process and enhance image quality during post-processing at the decoder.
In the encoder, optimization is applied to the SVM-CNN CU/PU algorithm by replacing the original interpolation filter with a bilinear interpolation filter to reduce computational load and save encoding time. However, simplifying fractional point estimation leads to image distortion. Hence, CNN is utilized for post-processing to improve the image, resulting in a reduction of BDBR% to 0.43% and an increase in TS% to 74.42%.
In the decoder, DenseNet/DAE three-channel CNN models are introduced to enhance decoded images, achieving a decrease in BDBR% to -5.96%.
In Chapter Four, we explore how to adjust the thresholds in the encoding algorithm to optimize the time-saving rate under the constraint of image quality improvement. By proportionally adjusting the thresholds for discriminative features such as SAD and RDO, we obtain spatial distributions for ΦSAD and ΦRDO concerning BDBR% and TS%. For more accurate predictions, experiments are conducted around BDBR% = 6.0%, resulting in relational equations between ΦSAD, ΦRDO, ZBDBR, and ZTS. We calculate the predicted optimal time-saving rate based on these equations. Finally, predictions are made based on the relationship curves between BDBR% and TS%. The results show an acceptable margin of error between predictions and experimental outcomes. Therefore, adjusting thresholds to optimize encoding calculations and predict BDBR% and TS% performance can further enhance overall efficiency in the future.
關鍵字(中) ★ 高效率視頻編碼
★ 畫面間預測
★ 卷積神經網路
★ 分散式影像編碼
★ 雙線性插值濾波器
★ 閥值
關鍵字(英) ★ High efficiency video coding
★ Inter Prediction
★ Convolutional Neural Networks
★ Distributed Video Coding
★ bilinear interpolation filter
★ thresholds
論文目次 章節目錄
論文摘要 vii
Abstract viii
誌謝 x
章節目錄 xi
第一章 緒論 1
1.1高效率視頻編碼(HEVC)標準介紹 1
1.1.1 高效率視頻編碼(HEVC)編碼流程 2
1.1.2 編碼單元(Coding Unit) 3
1.1.3 預測單元(Prediction Unit) 4
1.1.4 轉換單元(Transform Unit) 5
1.1.5 碼率失真代價函數(Rate Distortion Cost) 6
1.2 畫面間預測(Inter Prediction) 8
1.2.1 合併模式介紹(Merge Mode) 9
1.2.2 畫面間模式介紹(Inter Mode) 12
1.3 研究動機 18
1.3.1 論文架構 19
第二章 支持向量機與深度學習介紹 20
2.1支持向量機(Support Vector Machine) 20
2.2.1 類神經網路(Neural Network) 23
2.2.2 卷積神經網路(Convolutional Neural Network) 24
2.3相關文獻回顧 27
2.3.1 SVM 應用於 HEVC 畫面間編碼單元快速決策演算法 27
2.3.2 SVM-CNN 應用於 HEVC 畫面間編碼樹單元切割 36
第三章 以CNN為架構的分散式影像編碼(DVC)之探討 45
3.1分散式影像編碼(Distributed Video Coding) 45
3.1.1基於CNN的分散式影像編碼 47
3.2 DVC 編碼端探討 49
3.2.1以CNN改善 HEVC 插值器品質 49
3.2.2 SVM-CNN與CU/PU快速演算法 54
3.2.3 結合CNN-插值濾波器於SVM-CNN/CU-PU快速演算法 67
3.3 DVC 解碼端架構 72
3.3.1 DenseNet/Denoising Autoencoder應用於HEVC幀內後處理 72
3.3.2 HEVC 後處理畫面內/畫面間編碼比較 77
3.4 DVC編解碼端性能比較 81
3-4-2 DVC編解碼端整體效能分析 82
第四章 在影像品質限制下編碼時間最佳化之研究 86
4.1閥值參數設定 86
4.1.1 SAD閥值設定 87
4.1.2 RDO 閥值設定 90
4.2 在影像品質限制下(BDBR<6.0%)編碼時間最佳化 94
4.2.1 不同閥值組合下之BDBR(%)&TS(%)的空間分佈 95
4.2.2 BDBR% =6之下閥值間的關係方程式 99
4.2.3 BDBR(%)&TS(%)近似曲面方程式 102
4.3 不同影像品質下之編碼時間預測 107
4.3.1 近似曲面方程式之效能預測 107
第五章 結論與未來展望 111
參考文獻 113
參考文獻 [1]“Video coding for low bit rate communication, version 1,” ITU-T recommendation H.263, 1995.
[2] I. E. G. Richardson, H.264 and MPEG-4 Video Compression: Video Coding for Next-generation Multimedia. Aberdeen, U.K.: John Wiley & Sons,2003.
[3] Gary J. Sullivan, Jens-Rainer Ohm, Woo-Jin Han and Thomas Wiegand “Overview of the high efficiency video coding (HEVC) Standard,” in Proc. IEEE Transactions on circuits and systems for video technology,vol. 22, no. 12, pp. 1649-1668, Dec. 2012.
[4] L.Zhao, X. Guo, S. Lei, S. Ma and D. Zhao, “Simplified AMVP for high efficiency video coding,” in Proc. IEEE ICIP, pp. 1-4, 27-30 Nov.2012.
[5] Y. Ismail and S. El-etriby, “Fast diamond search algorithm for real time video coding,” in Proc. IEEE ICNC, pp. 729-733, Feb. 2012.
[6] J.K. Liu, “Efficient HEVC inter prediction using SVM,” Department of Communication Engineering National Central University, Taiwan 32054, R.O.C..
[7] D.H Yang “ SVM/CNN-based CTU Partition for HEVC Inter Prediction” Department of Communication Engineering National Central University,Taiwan 32054, R.O.C., Jan 2021.
[8] C.H Chen,“CNN-Based Post-Processing for HEVC Intra Prediction,”Department of Communication Engineering National Central University, Taiwan 32054, R.O.C., July 2020.
[9] P.H Tsui, “Post-Processing for HEVC Intra Prediction with ResNet algorithm” Department of Communication Engineering National CentralUniversity, Taiwan 32054, R.O.C., January 2022
[10] R. Puri and K. Ramchandran, “PRISM: A new robust video coding architecture based on distributed compression principles,”in Proceedings of the Allerton Conference on Communication, Control an d Computing, Allerton, IL, Oct. 2002.
[11] A. Aaron, R. Zhang, and B. Girod, “Wyner-Ziv Coding for Motion Video,” Asilomar Conference on Signals, Systems and Computers, Pacific Grove, USA, Nov. 2002.
[12] D. Slepian and J.K. Wolf, “Noiseless coding of correlated information sources,” IEEE Transactions on Information Theory, Vol. IT-19, July 1973, pp. 471–480.
[13] Wyner and J. Ziv, “The Rate-Distortion Function for Source Coding with Side Information at the Decoder”. IEEE Transactions on Information Theory, Vol. IT-22, Jan. 1976, pp. 1–10.
[14] T.Y Wei,“Efficient CU and PU Partitions for HEVC Inter Prediction”, Department of Communication Engineering National Central University,Taiwan 32054, R.O.C., January 2023.
[15] C,M Hang, “CNN-based HEVC interpolation filters”, Department of Communication Engineering National Central University, Taiwan 32054, R.O.C., July 2020.
[16] J. Kim, J.K. Lee, K.M. Lee, “Accurate Image Super-Resolution Using Very Deep Convolutional Networks”, The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 1646-1654.
[17] Y.C Chen,“Post-Processing for HEVC and VVC Intra Prediction With DenseNet / Denoising Autoencoder Algorithms”, Department of Communication Engineering National Central University,Taiwan 32054, R.O.C., January 2023.
[18] Y.D Tsai, “Research on Fast HEVC Inter Prediction Coding” National Central University, National Central University, Taiwan 32054, R.O.C., Jan 2019.
[19] S.J Cai, “Reduction of computation complexity for HEVC intra prediction with support vector machine,” National Central University, Master Thesis, Jun 2017.
指導教授 林銀議(Yin-Yi Lin) 審核日期 2024-1-22
推文 facebook   plurk   twitter   funp   google   live   udn   HD   myshare   reddit   netvibes   friend   youpush   delicious   baidu   
網路書籤 Google bookmarks   del.icio.us   hemidemi   myshare   

若有論文相關問題,請聯絡國立中央大學圖書館推廣服務組 TEL:(03)422-7151轉57407,或E-mail聯絡  - 隱私權政策聲明