Coefficient Position Coding Using Tree-Based Non-Zero Traversal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data compression techniques, such as zig-zag scanning, are inefficient when dealing with blocks containing only horizontal or vertical frequencies, as they visit a large number of zero coefficients, leading to inefficiencies in coding non-zero coefficients in image and video coders.
Innovation Solution
A method and apparatus that use a binary tree data structure to code the positions of non-zero coefficients, allowing for more flexible and efficient coding by traversing the tree to identify and encode non-zero coefficients, rather than relying on traditional scanning methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If zig-zag scanning method is used to code coefficient positions, then the coding process is simple and systematic, but it visits a large number of zero coefficients when blocks contain only horizontal or vertical frequencies, resulting in coding inefficiency
Solution Approach 1:
The patent segments the coefficient array into multiple sub-arrays based on the position of the first non-zero coefficient. Instead of scanning the entire array sequentially as in traditional zig-zag scanning, the array is divided into segments that can be processed independently, allowing the encoder to skip over regions containing only zero coefficients and focus only on segments containing non-zero coefficients, thereby improving coding efficiency.
Solution Approach 2:
The patent introduces a dynamic scanning approach where the scanning strategy adapts based on the content of the coefficient array. The encoder dynamically determines the starting position for scanning based on the location of the first non-zero coefficient, and adjusts the scanning pattern accordingly. This dynamic adaptation allows the system to optimize the scanning path for each specific block, avoiding unnecessary visits to zero coefficients while maintaining systematic processing.
2Adaptability or versatility
If traditional scanning methods are used, then the coding approach is uniform and easy to implement, but it lacks flexibility when dealing with different frequency distributions in coefficient blocks
Solution Approach 1:
The patent implements a dynamic scanning mechanism that automatically adapts to different frequency distributions in coefficient blocks. The encoder detects the position of the first non-zero coefficient and adjusts the scanning起始 position and pattern accordingly. This allows the same coding apparatus to efficiently handle various scenarios including blocks with horizontal frequencies, vertical frequencies, or mixed frequencies, providing high adaptability without requiring multiple specialized coding paths.
Solution Approach 2:
The patent changes the scanning parameters (starting position, scanning direction, segment boundaries) based on the actual content of the coefficient array. Instead of using fixed scanning parameters as in traditional methods, the system dynamically adjusts these parameters to match the frequency distribution characteristics of each block, thereby achieving versatility across different scenarios while maintaining a relatively simple implementation through parameter adjustment rather than structural complexity.
Data Source
AI summary
A method and apparatus for coding positions of data such as coefficients are described. In one embodiment, the method comprises coding data in a vector of data based on indications of whether the data is zero or non-zero identified using a tree data structure and producing a bit stream based on coded data.