簡易檢索 / 詳目顯示
| 研究生： |
林岱明 Tai-Ming Lin |
|---|---|
| 論文名稱： |
基於SVM與CNN的碼率控制方法應用於HEVC畫面內編碼 SVM/CNN-Based Approach to Rate Control in HEVC Intra Coding |
| 指導教授： | 林銀議 |
| 口試委員: | |
| 學位類別： |
碩士 Master |
| 系所名稱： |
資訊電機學院 - 通訊工程學系 Department of Communication Engineering |
| 論文出版年： | 2022 |
| 畢業學年度： | 110 |
| 語文別： | 中文 |
| 論文頁數： | 96 |
| 中文關鍵詞： | 高效率視訊編碼 、碼率控制 、畫面內預測 、支持向量機 、卷積神經網路 |
| 相關次數： | 點閱：18 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 |
在現今的社會，我們對於解析度的要求越來越高，為了因應我們所需高解析度的影像，高效率視訊編碼（HEVC）能比上一代的視訊編碼高出兩倍的壓縮率，這是因為在HEVC的壓縮技術中，使用了編碼單元、預測單元、轉換單元以及量化等方式。在網路傳輸方面，為了使傳輸的圖像有較低的失真量以及較好的效能，碼率控制就是在視訊編碼標準中實際使用的基本要素，在碼率控制中非常依賴位元速率以及編碼參數（量化參數、拉格朗日乘數）間的精確度。對於畫面間預測，可以根據先前編碼圖像的資訊來精確的更新參數以適應影片的內容，但對於畫面內預測卻是一大挑戰。在本論文中，為了能使畫面內預測有更精確的碼率控制方法，除了引用卷積神經網路外來預測每個編碼樹單元的參數外，也引用了支持向量機模型的特徵來進行訓練資料的分類，透過訓練資料的分類區分出平滑區塊以及複雜區塊後再進行卷積神經網路模型的訓練，使卷積神經網路模型預測更為精確。實驗結果表明，基於卷積神經網路以及支持向量機的方法，相較於高效率視訊編碼中HM 16.0的碼率控制方法的部分，在位元錯誤率方面降低了0.677%，且編碼效能也提升了0.78%。
In today's society, we have higher and higher requirements for resolution. In order to meet our needs for high-resolution images, High-Efficiency Video Coding(HEVC) can be twice as compressed as the previous generation of video coding. Because in the compression technology of HEVC, coding units, prediction units, transform units, and quantization methods are used. In terms of network transmission, in order to make the transmitted image have lower distortion and better performance, rate control is the basic element actually used in the video coding standard. Rate control scheme typically builds a model that characterizes the relationship between Bitrate and a coding parameter, e.g. quantization parameters and Lagrange multiplier(λ). For inter prediction, the parameters can be accurately updated based on the information of the previously coded image to adapt to the content of the video, however for intra prediction, it’s a challenge. In this paper, in order to enable a more accurate rate control method for intra prediction, in addition to quoting the convolutional neural network(CNN) to predict the parameters of each coding tree unit, the features of the support vector machine(SVM) model are also used to classify the training data. After distinguishing smooth and complex blocks through the classification of training data, making CNN model training is more accurate. The experimental results show that the method based on CNN and SVM reduces the bitrate error by 0.677% compared with The rate control method of HM 16.0 in HEVC and the coding performance is also increased by 0.78%.
[1] “Coding of audio-visual objects - Part 2: Visual,” in ISO/IEC 14496-2(MPEG-4 Visual Version 1), Apr. 1999.
[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] JCT-VC, “High efficiency video coding (HEVC) test model 15(HM15) encoder description,” JCTVC-Q1002, JCT-VC Meeting, Valencia, ES, Apr. 2014.
[4] G.J. Sullivan, J.R. Ohm, W.J. Han, T. Wiegand, “ Overview of the High Efficiency Video Coding (HEVC) Standard,” IEEE Trans. CSVT, vol. 22, no. 12, Dec. 2012.
[5] S.J. Cai, “Reduction of computation complexity for HEVC intra prediction with support vector machine,” National Central University, Master Thesis, Jun 2017.
[6] Y. Li, B. Li, D. Liu and Z. Chen, "A convolutional neural network-based approach to rate control in HEVC intra coding," 2017 IEEE Visual Communications and Image Processing (VCIP), 2017, pp. 1-4, doi: 10.1109/VCIP.2017.8305050.
[7] B. Li, H. Li, L. Li and J. Zhang, “λ domain rate control algorithm for high efficiency video coding”, IEEE Transactions on Image Processing, vol. 23, no. 9, pp. 3841-3854, Sept 2014.
[8] L. Li, B. Li, H. Li and C. W. Chen, "λ -Domain Optimal Bit Allocation Algorithm for High Efficiency Video Coding," in IEEE Transactions on Circuits and Systems for Video Technology, vol.28, no. 1, pp. 130-142, Jan. 2018, doi: 10.1109/TCSVT.2016.2598672.
