Efficient HEVC Scheme using Motion Type Categorization

Pallab Podder, Manoranjan Paul, Manzur Murshed

Research output: Book chapter/Published conference paperConference paperpeer-review

2 Citations (Scopus)
7 Downloads (Pure)

Abstract

High Efficiency Video Coding (HEVC) standard introduces a number of innovative tools which can reduce approximately 50% bit-rate compared to its predecessor H.264/AVC at the same perceptual video quality whereas the computational time has increased multiple times. To reduce the encoding time while preserving the expected video quality has become a real challenge today for video transmission and streaming especially using lowpowered devices. Motion estimation (ME) and motion compensation (MC) using variable-size blocks (i.e., intermodes) require 60-80% of total computational time. In this paper we propose a new efficient intermode selection technique based on phase correlation and incorporate into HEVC framework to predict ME and MC modes and perform faster intermode selection based on three dissimilar motion types in different videos. Instead of exploring all the modes exhaustively we select a subset of modes using motion type and the final mode is selected based on the Lagrangian cost function. The experimental results show that compared to HEVC the average computational time can be downscaled by 34% while providing the similar rate-distortion (RD) performance.
Original languageEnglish
Title of host publicationProceedings of the 10th International Conference on Emerging Networking EXperiments and Technologies (CoNEXT)
Place of PublicationUnited States
PublisherACM
Pages41-42
Number of pages2
DOIs
Publication statusPublished - 2014
Event10th International Conference on Emerging Networking EXperiments and Technologies (ACM CoNEXT 2014) - University of Technology, Sydney, Australia
Duration: 02 Dec 201405 Dec 2014

Conference

Conference10th International Conference on Emerging Networking EXperiments and Technologies (ACM CoNEXT 2014)
CountryAustralia
CitySydney
Period02/12/1405/12/14

Fingerprint Dive into the research topics of 'Efficient HEVC Scheme using Motion Type Categorization'. Together they form a unique fingerprint.

Cite this