| DC 欄位 |
值 |
語言 |
| DC.contributor | 通訊工程學系 | zh_TW |
| DC.creator | 王義品 | zh_TW |
| DC.creator | Yi-Pin Wang | en_US |
| dc.date.accessioned | 2024-1-22T07:39:07Z | |
| dc.date.available | 2024-1-22T07:39:07Z | |
| dc.date.issued | 2024 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=110523065 | |
| dc.contributor.department | 通訊工程學系 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 在這網路越來越普及且進步的時代下，隨著科技越來越
發達，使得人們對於品質有更高的要求及需求，在高解析度的影
像方面亦是如此，所以為了能夠有效的壓縮大量的資料下，H.266/
VVC 採用了許多更有效的技術，如方形矩形編碼樹單元、碼率
失真最佳化等等，但相隨而來的就是編碼計算複雜度的提升，而
本論文為結合近年來非常熱門的機器學習及深度學習，且應用於
VVC 畫面間預測中，一開始會先使用機器學習中的支持向量機
SVM 進行 CU 的劃分，劃分為後再使用深度學習 CNN 再更進一步
的劃分，最後再結合 CU-PU Decision 演算法把經過 SVM-CNN 區
分後的 Group，再一次地劃分。使得簡單的區塊不再需要去從頭計
算 RDO，經過以上種種且正確的分區。實驗結果與 VVC 相比，整體平均 BDBR 下降 2.03 百分比的情況下，整體編解碼時間節省可
以達到 49.59 百分比。 | zh_TW |
| dc.description.abstract | Since the development of technalogy , network becomes more
ubiquitous and advanced rapidly . Meanwhile , people are growing their demands and expectations for higher quality, and this trend extends to images with high resolution. To deal with the effective compression of massive data , VVC adopts various techniques , such as QTMT and RateDistortion Optimal . However , these precision processes also result in
high complexity in coding calculations . Then , our work aims at combining popular machine learning and deep learning , applying them to VVC inter prediction . At the beginning , we use machine learning method Support Vector Method on Coding Units(CUs) partition , and then employ the deep learning method Convolutional Neural Network for further refinement . Finally, integrating the CU-PU Decision algorithm and using it to determine the final partition for the groups defined by SVM-CNN allows simple blocks to skip the time-consuming Rate-Distortion Optimization (RDO). After the correct partitioning mentioned above, the experimental results show an average BDBR gain of -2.03%, with a total time-saving of 49.59% compared to VVC. | en_US |
| DC.subject | 多功能影像編碼 | zh_TW |
| DC.subject | 支持向量機 | zh_TW |
| DC.subject | 卷積神經網路 | zh_TW |
| DC.subject | 畫面間預測 | zh_TW |
| DC.subject | 編碼單元 | zh_TW |
| DC.subject | 快速深度決策 | zh_TW |
| DC.subject | 機器學習 | zh_TW |
| DC.subject | Versatile Video Coding | en_US |
| DC.subject | Support Vector Machine | en_US |
| DC.subject | Convolutional Neural Network | en_US |
| DC.subject | Inter Prediction | en_US |
| DC.subject | Coding Unit | en_US |
| DC.subject | Fast depth decision | en_US |
| DC.subject | Deep learning | en_US |
| DC.subject | Machine learning | en_US |
| DC.title | 快速 VVC 畫面間預測編碼之研究 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | Research on Fast VVC Inter Prediction Coding | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |