Low-Complexity Multiple Transform Selection Combining Multi-Type Tree Partition Algorithm for Versatile Video Coding

Area of Science:

• Digital video compression
• Image and video processing
• Computer vision

Background:

• Versatile Video Coding (VVC) offers improved compression over High-Efficiency Video Coding (HEVC).
• VVC's high computational complexity, particularly in Multiple Transform Selection (MTS), hinders real-time applications.
• Rate-Distortion Optimization (RDO) in MTS is a major contributor to encoding time.

Purpose of the Study:

• To develop a low-complexity MTS algorithm for VVC.
• To enable VVC for real-time sensor networks by reducing encoding time.
• To maintain high video compression efficiency while decreasing computational load.

Main Methods:

• Proposed a novel method to estimate Rate-Distortion (RD) cost for the last Coding Unit (CU), bypassing direct MTS RDO checks.
• Introduced an early termination strategy for MTS using information from neighboring Coding Units (CUs).
• Combined the optimized MTS with a multi-type tree partition algorithm.

Main Results:

• Achieved a 26.40% reduction in video encoding time compared to standard VVC.
• Observed a minimal increase of 0.13% in Bjøontegaard Delta Bitrate (BDBR), indicating preserved compression efficiency.
• Demonstrated the effectiveness of the proposed low-complexity MTS approach.

Conclusions:

• The proposed method significantly accelerates VVC encoding without substantial quality degradation.
• Optimized MTS is crucial for making VVC viable in real-time scenarios.
• This technique offers a practical solution for efficient video compression in resource-constrained environments.