.png){kind=icon}
.png){kind=icon}
Return
Swarm-Optimized Deblocking Filter for VVC: A Hybrid PSO–ML Artifact Reduction and Energy-Aware Video Coding
.png){kind=icon}
OA
AI
L
A
C
K
J
M
DOI：10.1016/j.icte.2026.05.012.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper proposes a Swarm-Optimized Deblocking Filter (SO-DF) to address artifact reduction in VVC. The proposed framework uses a combination of Particle Swarm Optimization (PSO) and a lightweight machine learning (ML) component to fine-tune filtering parameters. We have developed a system that integrates three innovative components: A Hybrid PSO-Gradient Descent Scheme; A CNN-Guided Swarm Initialization Strategy; and an Energy Aware Multi-Objective Fitness Function. The experimental evaluation of the JVET UHD test sequence demonstrated that SO-DF provides a 19.5% improvement in SSIM, with a 1.2% BD-rate savings and the ability to operate in real-time at 45 FPS.
Keywords:
Artifact Suppression
Deblocking Filter
Real-Time Video Processing
Swarm Optimization
VVC
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
4.2
Papers:
960
Citations:
2.5K
Organization
Cited Papers