{"title":"面向未来视频编码的基于QTBT结构的有效间变换方法","authors":"Liqiang Wang, Benben Niu, Yun He","doi":"10.1109/PCS.2018.8456266","DOIUrl":null,"url":null,"abstract":"Transform, a crucial module for hybrid video coding framework, has been selecting Discrete Cosine Transform (DCT) for several decades. Recently, Singular Value Decomposition (SVD) and Enhanced Multiple Transform (EMT) are proposed to improve transform efficiency. However, the perspectives of SVD and EMT are different. SVD enhances transform efficiency by utilizing the similarity of prediction block and inter residual block. EMT adopts some new sinusoidal transform cores to accommodate the larger prediction errors closer to the boundary of prediction unit. In this paper, the proposed method mainly has two key contributions. First, SVD and EMT are combined skillfully. Second, non-square SVD is newly introduced to the original algorithm. By extensive experiments, averages 1.07%, 1.06% and 0.65% BD-rate saving for Y, U and V are achieved compared to JEM5.0.1 with some coding tools off, up to 5.87%, 4.28% and 4.47%.","PeriodicalId":433667,"journal":{"name":"2018 Picture Coding Symposium (PCS)","volume":"50 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Effective Inter Transform Method Based on QTBT Structure for Future Video Coding\",\"authors\":\"Liqiang Wang, Benben Niu, Yun He\",\"doi\":\"10.1109/PCS.2018.8456266\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Transform, a crucial module for hybrid video coding framework, has been selecting Discrete Cosine Transform (DCT) for several decades. Recently, Singular Value Decomposition (SVD) and Enhanced Multiple Transform (EMT) are proposed to improve transform efficiency. However, the perspectives of SVD and EMT are different. SVD enhances transform efficiency by utilizing the similarity of prediction block and inter residual block. EMT adopts some new sinusoidal transform cores to accommodate the larger prediction errors closer to the boundary of prediction unit. In this paper, the proposed method mainly has two key contributions. First, SVD and EMT are combined skillfully. Second, non-square SVD is newly introduced to the original algorithm. By extensive experiments, averages 1.07%, 1.06% and 0.65% BD-rate saving for Y, U and V are achieved compared to JEM5.0.1 with some coding tools off, up to 5.87%, 4.28% and 4.47%.\",\"PeriodicalId\":433667,\"journal\":{\"name\":\"2018 Picture Coding Symposium (PCS)\",\"volume\":\"50 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2018-06-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2018 Picture Coding Symposium (PCS)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/PCS.2018.8456266\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2018 Picture Coding Symposium (PCS)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/PCS.2018.8456266","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Effective Inter Transform Method Based on QTBT Structure for Future Video Coding
Transform, a crucial module for hybrid video coding framework, has been selecting Discrete Cosine Transform (DCT) for several decades. Recently, Singular Value Decomposition (SVD) and Enhanced Multiple Transform (EMT) are proposed to improve transform efficiency. However, the perspectives of SVD and EMT are different. SVD enhances transform efficiency by utilizing the similarity of prediction block and inter residual block. EMT adopts some new sinusoidal transform cores to accommodate the larger prediction errors closer to the boundary of prediction unit. In this paper, the proposed method mainly has two key contributions. First, SVD and EMT are combined skillfully. Second, non-square SVD is newly introduced to the original algorithm. By extensive experiments, averages 1.07%, 1.06% and 0.65% BD-rate saving for Y, U and V are achieved compared to JEM5.0.1 with some coding tools off, up to 5.87%, 4.28% and 4.47%.