Adaptive Context Model Selection with Zigzag Scan
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The adaptive scan scheme in HEVC introduces substantial computational complexity and additional memory requirements while achieving only a negligible performance gain in reducing bandwidth requirements for transform processing in video compression.
Innovation Solution
Implementing adaptive context model selection using a zigzag scan pattern for coding and decoding, which reduces inefficiencies by eliminating the need to track counts of coded significant transform coefficients and perform branch operations, thereby simplifying scan selections in significance map processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If an adaptive scan scheme is implemented in HEVC, then scan selection adaptability is improved, but device complexity increases due to tracking counts and performing branch operations
Solution Approach 1:
The significance map is divided into multiple sub-blocks, and a separate context model is selected for each sub-block based on local characteristics. This segmentation allows adaptive scanning without requiring complex global tracking, as each sub-block can be processed independently with its own context model selection.
Solution Approach 2:
Context models are pre-defined and prepared in advance for different scanning patterns. During encoding/decoding, the appropriate pre-prepared context model is selected based on local significance map characteristics, eliminating the need for complex real-time tracking and branch operations while maintaining adaptability.
2Adaptability or versatility
If an adaptive scan scheme is implemented in HEVC, then scan selection adaptability is improved, but device complexity increases due to additional memory requirements
Solution Approach 1:
By segmenting the significance map into sub-blocks, each with its own context model, the memory requirement is distributed across smaller units rather than requiring large memory for global adaptive scanning tables. This reduces peak memory requirements while maintaining adaptability.
Solution Approach 2:
Instead of storing complex adaptive scan patterns and tracking data in memory, the patent uses replicated context model tables that can be efficiently stored and reused. Multiple copies of simplified context model data replace the need for large memory structures required by traditional adaptive scanning.
3Productivity
If adaptive context model selection with zigzag scan is implemented, then coding efficiency is improved by reducing computational overhead, but adaptability may be reduced compared to full adaptive scanning
Solution Approach 1:
The patent applies different context models to different local regions (sub-blocks) of the significance map based on their specific characteristics. This local differentiation maintains adaptability where needed while using a fixed zigzag scan pattern globally, achieving a balance between coding efficiency and adaptability.
Solution Approach 2:
The fixed zigzag scan pattern serves multiple functions: it provides a consistent baseline scanning order for all blocks and works effectively with the context model selection mechanism. This universal approach eliminates the need for complex adaptive scan pattern switching while maintaining good coding performance through context-aware probability modeling.
Data Source
AI summary
There is a coding. The coding may include preparing video compression data based on source pictures utilizing a processor. The preparing may include processing a generated transform unit, including generating a significance map having a significance map array with y-x locations corresponding to the transform array. The generating may include scanning, utilizing a zigzag scanning pattern, a plurality of significance map elements in the significance map array. The generating may also include determining, utilizing the zigzag scanning pattern, a context model for coding a significance map element of the plurality of significance map elements based on a value associated with at least one coded neighbor significance map element of the significance map element in the significance map array. There is also a decoding including processing video compression data which is generated in the coding.