A Low Complexity Spatial Texture Measurement for Fast CU Size Decision and Mode Decision in HEVC
Date Issued
2014
Date
2014
Author(s)
Wu, Tai-Ting
Abstract
Abstract— High Efficiency Video Coding (HEVC) [1] is a new video compression standard which reduces bitrate by half compared with H.264/AVC. But the computational complexity increases due to the quadtree based structure, which contributes to significant coding efficient improvement. As a result, reducing this complexity and maintaining negligible loss in bitrate and peak signal-to-noise (PSNR) simultaneously is a must. In this work, we design a fast mode decision scheme for inter and intra prediction in HEVC based on the texture complexity. We show that the relation between texture complexity and mode decision, then use this relation to speed up the encoding process.
The proposed scheme provides a decrease of up to 53.4% and by averagely 46.5% for intra coding, with negligible loss in bitrate and PSNR. On the other hand, the proposed scheme for inter coding also provides a decrease of up to 38.6% and by averagely 31.2% with negligible loss in bitrate and PSNR.
Subjects
高效率視訊編碼
影格內編碼
影格間編碼
影片內容複雜度
四元樹架構
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-103-R01944035-1.pdf
Size
23.32 KB
Format
Adobe PDF
Checksum
(MD5):96179cf723387852461c475f64a58de2