Quantization Parameter Cascading for Surveillance Video Coding Considering All Inter Reference Frames

Area of Science:

• Video Compression
• Digital Signal Processing
• Computer Vision

Background:

• Video surveillance systems are widespread, requiring efficient video coding for transmission and storage.
• Low-delay (LD) configurations of High Efficiency Video Coding (HEVC) are crucial for real-time surveillance video transmission.
• Current Quantization Parameter Cascading (QPC) in HEVC test models (HM) lacks optimization due to unaddressed reference dependency.

Purpose of the Study:

• To propose an efficient Quantization Parameter Cascading (QPC) technique for surveillance video coding.
• To enhance the coding efficiency of HEVC's low-delay (LD) configuration by optimizing QP selection.
• To address the limitations of existing QPC methods in HEVC by incorporating inter-reference frame dependencies.

Main Methods:

• Developed a novel QPC technique, termed QPC-SV, specifically for surveillance video coding.
• Integrated consideration of all inter-reference frames within the HEVC LD configuration.
• Evaluated QPC-SV against the default QPC in the HEVC test model (HM).

Main Results:

• The proposed QPC-SV technique demonstrates significant rate-distortion performance gains.
• Achieved average BD-rate improvements of -9.35% for LDP and -9.76% for LDB configurations.
• The method effectively leverages reference frame information for enhanced coding efficiency.

Conclusions:

• QPC-SV offers a superior approach to Quantization Parameter Cascading for HEVC LD surveillance video coding.
• The proposed technique provides substantial improvements in coding efficiency compared to existing methods.
• Optimizing QPC by considering inter-reference frames is vital for effective surveillance video compression.