HDR视频编码的自适应感知量化器：HEVC Main 10 Profile下的适应性PQ解决方案

84 浏览量更新于2024-08-26 1 收藏 394KB PDF 举报

本文档由Joint Collaborative Team on Video Coding (JCT-VC) 在2016年2月9日的第23次会议上提出，地点位于美国圣地亚哥。讨论的主题是“CE2-related: Adaptive PQ - 一种基于HEVC Main 10 Profile的自适应感知量化器在HDR视频编码中的应用”。该文档作为输入文档提交给了JCT-VC，其目标是解决高动态范围（HDR）视频编码中感知量化器（PQ）的非适应性问题。传统的PQ方法使用固定的映射函数将亮度映射到luma（Y分量），这在处理HDR内容时可能导致质量不均或不适合不同亮度场景。为了解决这个问题，作者C. Jung、S. Yu、P. Ke来自西安电子科技大学和赵志坤教授团队，以及M. Li和P. Wu来自南京ZTE研究中心提出了自适应感知量化器（Adaptive PQ）。这个新型的量化器引入了根据HDR内容的亮度信息计算出的比例因子，从而实现了对亮度的自适应映射。自适应PQ通过动态调整亮度到luma的转换，能够更好地捕捉HDR视频中的细节和动态范围变化，提高了编码效率和视觉质量。实验结果显示，与传统的固定PQ相比，自适应PQ在HDR视频编码中表现出显著的优势，包括更平滑的图像质量和更高的主观观感满意度。文章详细介绍了自适应PQ的工作原理、设计思路以及实施方法，并展示了实际的实验结果，以证明其在HDR视频编码中的实用性。此外，联系信息提供了进一步交流和合作的渠道，包括作者的联系方式和机构背景，表明这项技术来自西安电子科技大学和ZTE Corporation的联合研究。这篇研究论文提出了一个创新的解决方案，旨在提升HDR视频编码的性能，对于理解HEVC Main 10 Profile在HDR内容处理中的优化策略具有重要意义，也为后续的视频编码标准和实践应用提供了有价值的技术参考。

Page: 1 Date Saved: 2016-02-09

Joint Collaborative Team on Video Coding (JCT-VC)

of ITU-T SG 16 WP 3 and ISO/IEC JTC 1/SC 29/WG 11

23rd Meeting: San Diego, USA, 19–26 February 2016

Document: JCTVC-W0071

Title:

CE2-related: Adaptive PQ: Adaptive Perceptual Quantizer for HDR video Coding

with HEVC Main 10 Profile

Status: Input Document to JCT-VC

Purpose: Proposal

Author(s) or

Contact(s):

C. Jung, S. Yu, P. Ke

Xidian University

2 South Taibai Road

Xian 710071, China

M. Li, P. Wu

No.6 Huashendadao Road, ZTE R&D

Center, Nanjing 210012, China

Tel:

Email:

+86-29-88202654

zhengzk@xidian.edu.cn

Source: Xidian University, ZTE Corporation

_____________________________

Abstract

Proposed is an adaptive transform function based perceptual quantizer (PQ) for HDR videos, called

adaptive PQ. Adaptive PQ is to solve the problem that PQ is not adaptive to HDR content because of its

using a fixed mapping function from luminance to luma. By introducing a ratio factor derived from the

luminance information of HDR content, adaptive PQ maps luminance to luma adaptively according to the

content. Experimental results demonstrate that adaptive PQ achieves average bit-rate reductions of 3.51%

in tPSNR XYZ and 5.51% in PSNR_DE over PQ.

1 Introduction

With the rapid advances in the HDR technology, efficient storage and transmission of such data are

crucial for HDR value chain. Several methods have been proposed to handle this challenge. First,

backward compatible approaches [1][2] have been proposed to compress HDR videos. In the backward

compatible approaches, HDR videos are divided into two layers: base layer of low dynamic range (LDR)

data and enhancement layer of residual data. LDR data is obtained by tone mapping on the HDR videos,

while residual data is used to reconstruct the HDR videos. Second, lossless HDR image formats such as

Radiance [3], OpenEXR [4], and LogLuv [5] have been proposed to store floating-point colorimetric

HDR data [6][7]. Although lossless formats accurately record luminance in the scene, floating-point data

needs huge memory to store. Moreover, compared with integer, floating-point data is inefficient to

encode. In general, the input data is integer in digital image and video processing. To widely use HDR

contents, floating-point data of HDR pixels should be transformed into integer data. Aiming at solving the

problem, several perceptual transfer functions (TFs) for HDR contents have been proposed to convert

floating-point luminance to integer data [8]-[13]. In [9], the integer data is luma which means the

perceptually encoded and quantized luminance.

下载后可阅读完整内容，剩余6页未读，立即下载

weixin_38734269

粉丝: 3
资源: 930

HDR视频编码的自适应感知量化器：HEVC Main 10 Profile下的适应性PQ解决方案

PQ-EOTF-PPT.pdf

使用感知块合并优化HDR视频中的HEVC Main 10编码

kvazaar:开源HEVC编码器

Optimization of HEVC Main 10 coding in HDR videos Based on Disorderly Concealment Effect

FFmpeg Hi:具有HEVC（8/10位）和HE-AAC的FFmpeg-开源

基于块级JND的视频编码自适应感知量化算法

3D-HEVC的感知非对称视频编码

ShanaEncoder 4.8中文版：高效HEVC(H265)视频压缩软件

平衡速度与性能：面向HEVC的自适应快速运动估计算法

最新资源