H.264 SVC Complexity Reduction Based on Likelihood Mode Decision.

Area of Science:

• Computer Science
• Digital Signal Processing
• Video Compression

Background:

• H.264 Advanced Video Coding (AVC) has been extended to Scalable Video Coding (SVC) to support diverse devices and user needs.
• SVC enables content scaling across various parameters like resolution, frame rate, and quality for heterogeneous networks and bandwidths.
• The scaling process in SVC increases encoding time and computational complexity, particularly during mode selection.

Purpose of the Study:

• To propose a fast mode decision algorithm to reduce encoding time and computational complexity in SVC.
• To evaluate the proposed algorithm's performance in both temporal and spatial scaling scenarios.

Main Methods:

• A novel Likelihood Mode Decision (LMD) algorithm was developed for fast mode selection in SVC.
• The LMD algorithm was evaluated against previous fast mode decision algorithms using key performance metrics.

Main Results:

• The LMD algorithm demonstrated significant time savings, achieving 66.65% reduction compared to the full search method.
• A minor detriment of 0.05% in Peak Signal-to-Noise Ratio (PSNR) and a 0.17% increment in bit rate were observed.
• LMD outperformed existing fast mode decision algorithms in terms of efficiency and effectiveness.

Conclusions:

• The proposed Likelihood Mode Decision (LMD) algorithm effectively reduces encoding time and computational complexity in Scalable Video Coding (SVC).
• LMD offers a favorable trade-off between encoding efficiency and video quality metrics like PSNR and bit rate.
• This algorithm presents a viable solution for optimizing SVC performance across various electronic devices and network conditions.