Efficient Multi-Strategy Intra Prediction for Quality Scalable High Efficiency Video Coding.

Area of Science:

• Video Compression Technologies
• Digital Signal Processing
• Computer Vision

Background:

• High Efficiency Video Coding (HEVC) is complex, and its extension, Scalable HEVC (SHVC), further increases computational demands.
• Intra prediction in SHVC, particularly for quality enhancement layers, involves extensive mode selection and depth-level analysis, leading to high complexity.

Purpose of the Study:

• To propose a novel intra prediction scheme to significantly speed up the enhancement layer intra-coding in Quality SHVC.
• To reduce the computational complexity associated with mode decision and depth selection in SHVC intra-coding.

Main Methods:

• A multi-step, multi-strategy approach combining spatial and inter-layer correlations to predict probable depth levels.
• Utilizing residual distribution analysis (skewness, kurtosis) to conditionally skip complex Inter-Layer Reference (ILR) mode comparisons.
• Employing local monotonicity of Hadamard costs and hypothesis testing for early termination of depth selection and mode comparisons.

Main Results:

• The proposed scheme achieves an average speedup gain of over 75% on test video sequences.
• Maintained coding efficiency comparable to existing methods despite significant complexity reduction.

Conclusions:

• The developed scheme effectively minimizes depth selections and reduces mode decision complexity in a hierarchical manner.
• This approach offers a practical solution for accelerating SHVC enhancement layer intra-coding without sacrificing compression performance.