Effective CU size decision for HEVC intracoding.

Area of Science:

• Video Compression
• Digital Signal Processing
• Computer Vision

Background:

• High Efficiency Video Coding (HEVC) utilizes a tree-structured coding unit (CU) for recursive splitting.
• HEVC supports numerous intraprediction modes, enhancing coding efficiency but increasing computational complexity due to exhaustive search.
• The standard HEVC Test Model (HM) faces high computational demands from its intraprediction process.

Purpose of the Study:

• To propose a fast CU size decision algorithm for HEVC intracoding.
• To reduce the computational complexity associated with HEVC intraprediction.
• To accelerate the video coding process without significant quality degradation.

Main Methods:

• Developed an early CU size determination method using adaptive thresholds based on texture homogeneity.
• Introduced a novel bypass strategy for intraprediction on large CUs, considering texture and neighboring coding information.
• Implemented and evaluated the algorithm within the HEVC Test Model (HM).

Main Results:

• Achieved a computational complexity reduction of up to 67%.
• Observed a minimal Peak Signal-to-Noise Ratio (PSNR) loss of only 0.06 dB.
• Reported a slight bit rate increase of 1.08% compared to the original HM coding.

Conclusions:

• The proposed fast CU size decision algorithm effectively reduces HEVC intracoding complexity.
• The algorithm offers a practical solution for accelerating video encoding while maintaining high video quality.
• This approach balances computational efficiency and coding performance in HEVC.