基于超像素的视频编码帧间预测

Md. Eimran Hossain Eimon, Md. Zahirul Islam, Md. Shahid Uz Zaman, Md. Al Mehedi Hasan, Boshir Ahmed
{"title":"基于超像素的视频编码帧间预测","authors":"Md. Eimran Hossain Eimon, Md. Zahirul Islam, Md. Shahid Uz Zaman, Md. Al Mehedi Hasan, Boshir Ahmed","doi":"10.1109/ICAEE48663.2019.8975508","DOIUrl":null,"url":null,"abstract":"Traditional video codec uses the block-based approach for motion estimation and compensation, which is unable to capture the true underlying motion. Blocks consisting the boundary of moving object need more bit for better prediction of the current frame because real-world objects are not block-based and their motion is not translational. One of the main goal of the current standard (HEVC) is to minimize the prediction error energy by using quad-tree splitting method. But if motion discontinuity exits, quad-tree based splitting method uses a large amount of motion bit-rate and also introduces a potential coding overhead in the video encoder and decoder. In this paper, we propose a segmentation-based method of using arbitrary shape superpixel and affine motion registration technique for reducing the prediction error energy between the current and reference frame. This method uses the already encoded reference frame for predicting the current frame. The experimental result shows that good subjective quality with PSNR up to 27.71dB can be obtained using this approach.","PeriodicalId":138634,"journal":{"name":"2019 5th International Conference on Advances in Electrical Engineering (ICAEE)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Superpixel Based Inter-Frame Prediction for Video Coding\",\"authors\":\"Md. Eimran Hossain Eimon, Md. Zahirul Islam, Md. Shahid Uz Zaman, Md. Al Mehedi Hasan, Boshir Ahmed\",\"doi\":\"10.1109/ICAEE48663.2019.8975508\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Traditional video codec uses the block-based approach for motion estimation and compensation, which is unable to capture the true underlying motion. Blocks consisting the boundary of moving object need more bit for better prediction of the current frame because real-world objects are not block-based and their motion is not translational. One of the main goal of the current standard (HEVC) is to minimize the prediction error energy by using quad-tree splitting method. But if motion discontinuity exits, quad-tree based splitting method uses a large amount of motion bit-rate and also introduces a potential coding overhead in the video encoder and decoder. In this paper, we propose a segmentation-based method of using arbitrary shape superpixel and affine motion registration technique for reducing the prediction error energy between the current and reference frame. This method uses the already encoded reference frame for predicting the current frame. The experimental result shows that good subjective quality with PSNR up to 27.71dB can be obtained using this approach.\",\"PeriodicalId\":138634,\"journal\":{\"name\":\"2019 5th International Conference on Advances in Electrical Engineering (ICAEE)\",\"volume\":\"7 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2019-09-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2019 5th International Conference on Advances in Electrical Engineering (ICAEE)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/ICAEE48663.2019.8975508\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2019 5th International Conference on Advances in Electrical Engineering (ICAEE)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICAEE48663.2019.8975508","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

摘要

传统的视频编解码器采用基于块的方法进行运动估计和补偿,无法捕捉到真实的底层运动。包含移动物体边界的块需要更多的比特来更好地预测当前帧,因为现实世界的物体不是基于块的,它们的运动不是平移的。当前标准(HEVC)的主要目标之一是利用四叉树分裂方法使预测误差能量最小化。但是如果存在运动不连续,基于四叉树的分割方法使用了大量的运动比特率,并且在视频编码器和解码器中引入了潜在的编码开销。本文提出了一种基于分割的方法,利用任意形状的超像素和仿射运动配准技术来降低当前帧和参考帧之间的预测误差能量。该方法使用已经编码的参考帧来预测当前帧。实验结果表明,该方法可获得较好的主观质量,PSNR高达27.71dB。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Superpixel Based Inter-Frame Prediction for Video Coding
Traditional video codec uses the block-based approach for motion estimation and compensation, which is unable to capture the true underlying motion. Blocks consisting the boundary of moving object need more bit for better prediction of the current frame because real-world objects are not block-based and their motion is not translational. One of the main goal of the current standard (HEVC) is to minimize the prediction error energy by using quad-tree splitting method. But if motion discontinuity exits, quad-tree based splitting method uses a large amount of motion bit-rate and also introduces a potential coding overhead in the video encoder and decoder. In this paper, we propose a segmentation-based method of using arbitrary shape superpixel and affine motion registration technique for reducing the prediction error energy between the current and reference frame. This method uses the already encoded reference frame for predicting the current frame. The experimental result shows that good subjective quality with PSNR up to 27.71dB can be obtained using this approach.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信