Coefficient Group Re-positioning for Video Encoding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for processing next-generation video content with high spatial resolution, high frame rate, and high dimensionality require significant increases in memory storage, memory access rate, and processing power, necessitating more efficient coding tools.
Innovation Solution
A method for encoding and decoding transformed residual signals involves dividing quantized coefficient blocks into coefficient groups, re-positioning non-zero coefficients, and performing entropy encoding and decoding in specific orders to enhance efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If next-generation video content with high spatial resolution, high frame rate, and high dimensionality is processed using conventional methods, then video quality and resolution are improved, but memory storage requirements, memory access rate, and processing power requirements increase drastically
Solution Approach 1:
The transform coefficient block is divided into multiple sub-blocks, and each sub-block is further divided into coefficient groups (CGs) of predetermined size (e.g., 2×2). This hierarchical segmentation allows the encoder to process and encode coefficients in smaller units, reducing the computational burden and memory requirements while maintaining high video resolution.
Solution Approach 2:
The encoder performs preliminary organization of coefficient groups by identifying the last CG containing non-zero coefficients and re-positioning non-zero coefficients from the first CG to the second CG before entropy encoding. This preliminary arrangement optimizes the encoding process by reducing the number of zero coefficients that need to be processed, thereby decreasing processing power requirements.
2Productivity
If conventional entropy encoding is performed on quantized coefficient blocks without re-positioning, then encoding process is simpler, but the amount of data requiring encoding increases
Solution Approach 1:
The method extracts non-zero coefficients from the first CG (which contains fewer non-zero coefficients) and re-positions them into the second CG (which contains more non-zero coefficients). This extraction and re-positioning consolidates non-zero coefficients, reducing the total number of coefficients that require entropy encoding and thereby reducing the amount of residual signal data.
Solution Approach 2:
The encoder changes the positional parameters of non-zero coefficients by re-positioning them from the first CG to the second CG. This parameter change optimizes the distribution of non-zero coefficients, improving entropy encoding efficiency by reducing the number of zero coefficients that need to be processed during encoding.
3Manufacturing precision
If coefficient groups are re-positioned based on inverse-scanning order, then encoding precision is improved, but processing complexity increases
Solution Approach 1:
The method dynamically determines which CG is the first CG and which is the second CG based on the inverse-scanning order and the distribution of non-zero coefficients. This dynamic identification and re-positioning approach allows the encoder to adapt to different coefficient distributions while maintaining a relatively simple processing framework, balancing encoding precision with processing complexity.
Data Source
AI summary
The present invention provides methods and devices for re-positioning significant coefficient in image encoding/decoding process. A method according to an embodiment of the present invention comprises relocating the non-zero coefficient of the first coefficient group to a preceding coefficient group of the last coefficient group according to the inverse-scanning order based on a number of non-zero coefficients of the last coefficient group and a number of zero coefficients at positions starting from the last non-zero coefficients within the preceding coefficient group, thereby an amount of residual signals may be saved.