基於支持向量機之HEVC低運算複雜度畫面內預測演算法研究;Low Computational Complexity Algorithm for HEVC Intra Prediction with Support Vector Machine

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Communication Engineering > Electronic Thesis & Dissertation > Item 987654321/68675

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/68675

Title:	基於支持向量機之HEVC低運算複雜度畫面內預測演算法研究;Low Computational Complexity Algorithm for HEVC Intra Prediction with Support Vector Machine
Authors:	黃上恩;Huang,Shang-en
Contributors:	通訊工程學系
Keywords:	HEVC;編碼單元;畫面內預測;支持向量機;HEVC;Coding Unit;Intra Prediction;SVM
Date:	2015-07-30
Issue Date:	2015-09-23 14:00:46 (UTC+8)
Publisher:	國立中央大學
Abstract:	最新視訊編碼標準為高效率視訊編碼(High Efficiency Video Coding, HEVC)，HEVC比起先前的視訊編碼標準H.264/AVC編碼效率更佳。HEVC的編碼單元(Coding Unit , CU)採用了四分樹(Quad-Tree)的編碼架構增加其編碼效能，在畫面內預測中，為了使預測更精準，使用了35個模式來進行預測，同時也大幅的增加其編碼的複雜度。因此本篇論文提出一個應用於畫面內編碼的CU深度快速決策演算法，擷取三種原始畫面的資訊以及空間上的相關性做為特徵(Feature)，其中包含了變異數、低頻交流值以及鄰近編碼單元的深度資訊，依照輸入特徵給予支持向量機(Support vector machine, SVM)所預測出的結果，判斷當前CU是否往下進行切割；於畫面內預測中，則是使用索爾運算子(Sobel)來判斷當前畫面預測單元(Prediction Unit,PU)的主要方向以及鄰近幾個方向做為我們約略模式決策(Rough Mode Decision, RMD)的候選模式，而且我們也減少了RDO(Rate-Distortion Optimization)的候選模式個數以達到減省時間的效果。由實驗結果可知，我們所提出的演算法在維持一定影像品質之下，平均能夠節省48.3%的整體編碼時間。;High efficiency video coding (HEVC) is the state-of-the-art video coding standard. HEVC is better than the previous video coding standard H.264/AVC on the coding efficient. The coding unit(CU) of HEVC uses a Quadtree-based coding structure to increase coding performance. In order to predict more accurately, using 35 prediction modes in intra prediction. Simultaneously, it also increases its coding complexity. Hence, in this thesis, we proposed an SVM based fast intra CU depth decision algorithm is proposed to reduce the computational complexity. It is convenient to develop the criterion of early CU splitting and termination by applying SVM with features extracted from spatial domain, frequency domain, including Variance, low-frequency AC value, and neighboring CU depth. After extracting the features, SVM predict the current CU will be split or be terminated. In intra prediction, we use sobel operator to distinguish the direction of PU(Prediction Unit) and the neighbor mode to be the candidates of rough mode decision and we also reduce the candidates of rate-distortion optimization. The experimental results show that our proposed algorithm achieves 48.3% encoding time saving on average without significant degradation of coding performance.
Appears in Collections:	[Graduate Institute of Communication Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	436	View/Open

社群 sharing

Loading...