View : 1088 Download: 0

Full metadata record

DC Field Value Language
dc.contributor.advisor강제원-
dc.contributor.author류수경-
dc.creator류수경-
dc.date.accessioned2019-05-07T16:30:10Z-
dc.date.available2019-05-07T16:30:10Z-
dc.date.issued2018-
dc.identifier.otherOAK-000000147624-
dc.identifier.urihttp://dcollection.ewha.ac.kr/common/orgView/000000147624en_US
dc.identifier.urihttps://dspace.ewha.ac.kr/handle/2015.oak/249820-
dc.description.abstract본 논문에서는, 임의 결정 트리의 앙상블 모델인 랜덤 포레스트를 이용한 머신 러닝 기반의 고속 화면 내 예측 모드 결정 기법에 대해 제안한다. 랜덤 포레스트는 예측 단위(Prediction Unit)로부터 화면 내 예측 모드를 예측한다. 예측한 모드를 이용하여 율-왜곡 최적화 과정을 하는 화면 내 예측 후보 모드 수를 줄임으로써, 부호화 시간을 단축 시킨다. 이를 위해, 노드마다 파라미터화 된 분기 함수를 포함하는 임의 결정 트리를 통해 방향성을 갖는 블록 기반의 특징(feature)을 학습한다. 학습 feature는 블록의 방향성을 반영하는 4개의 픽셀 값을 이용해 구하였고, 이를 통해 빠르고 효율적인 결과를 얻을 수 있었다. 제안 기법과 일반적인 비디오 부호화 표준 체계를 결합하기 위해서, 랜덤 포레스트를 통해 도출한 화면 내 예측 모드를 율-왜곡 최적화 과정에 도달하기 전 기존 코덱의 후보 모드 군에 포함시킨다. 제안 기법은 현재 비디오 압축 표준인 High Efficiency Video Coding (HEVC)의 참조 소프트웨어인 HM과 Joint Exploration Model(JEM)에 구현되었다. 실험 결과는 제안 기법이 참조 소프트웨어 모델보다 낮은 부호화 손실로 복잡도를 낮추었음을 보여준다.;In this thesis, we propose a machine learning-based fast intra-prediction mode decision algorithm, using random forest that is an ensemble model of randomized decision trees. The random forest is used to estimate an intra-prediction mode from a prediction unit and to reduce encoding time significantly by avoiding the intensive Rate-Distortion optimization of a number of intra-prediction modes. To this aim, we develop a randomized tree model including parameterized split functions at nodes to learn directional block-based features. The feature uses only four pixels reflecting directional property of a block, and, thus the evaluation is fast and efficient. To integrate the proposed technique into the conventional video coding standard frameworks, the intra-prediction mode derived from the proposed technique, called an inferred mode (IM), is used to shrink the pool of the candidate modes before carrying out the Rate-Distortion(R-D) optimization. The proposed technique is implemented into the High Efficiency Video Coding Test Model (HM) reference software of the state-of-the-art video coding standard and Joint Exploration Model (JEM) reference software, by integrating the random forest trained off-line into the codecs. Experimental results demonstrate that the proposed technique achieves significant encoding time reduction with only slight coding loss as compared the reference software models.-
dc.description.tableofcontentsI. Introduction 1 A. Study Object 1 B. Thesis Order 4 II. Related Work 5 A. Fast Intra Mode Decision 5 B. Machine Learning-Based Fast Coding Techniques 7 III. Preliminary of the Standard Random Forest 9 A. Decision Tree Training 11 B. Testing 12 IV. The Proposed Technique 13 A. Intra-Prediction Mode Estimation Using Random Forest 13 1. Directional Block-based Features 13 2. Optimization of a Split Function 14 3. Training a Prediction Unit with a Random Forest 15 4. Decision of a Prediction Mode with a Random Forest 18 B. Implementation to a Codec 22 V. Experimental Results 25 A. Coding Configurations and Test Sequence 25 B. Coding Performance Evaluation and Analysis 28 1. Performance Evaluation in HEVC reference software 28 2. Performance Evaluation in JEM software 32 C. Performance Evaluation with Various Configurations in Random Forest 34 VI. Conclusion 37 Reference 38 Abstract (in Korean) 44-
dc.formatapplication/pdf-
dc.format.extent1767052 bytes-
dc.languageeng-
dc.publisher이화여자대학교 대학원-
dc.subject.ddc000-
dc.titleMachine Learning-Based Fast Angular Prediction Mode Decision Technique in Video Coding-
dc.typeMaster's Thesis-
dc.format.pageiv, 44 p.-
dc.identifier.thesisdegreeMaster-
dc.identifier.major대학원 전자전기공학과-
dc.date.awarded2018. 2-
Appears in Collections:
일반대학원 > 전자전기공학과 > Theses_Master
Files in This Item:
There are no files associated with this item.
Export
RIS (EndNote)
XLS (Excel)
XML


qrcode

BROWSE