Low-Complexity Transform Kernel Sequence for Video Residual Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding/decoding technologies face high complexity and computational load due to the need for efficient processing of high-resolution and high-quality video, particularly in transforming residual blocks for encoding and decoding.
Innovation Solution
The method employs a two-step transform process using a low-complexity first transform kernel, such as Walsh-Hadamard or 2-square, followed by a sparse transform kernel, to reduce computational complexity and improve encoding/decoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a conventional transform kernel is used for transforming residual blocks in high-resolution video encoding, then transform accuracy is maintained, but computational complexity increases significantly
Solution Approach 1:
The transform process is segmented into two sequential steps: first applying a low-complexity transform kernel to the residual block, then applying a sparse transform kernel to the transformed coefficients. This segmentation allows the system to break down the computationally intensive single transform into manageable stages, reducing overall complexity while preserving accuracy
Solution Approach 2:
The first transform kernel acts as an intermediary that prepares the residual block for the second transform. By introducing this intermediate transformation step, the system achieves a bridge between the original residual data and the final transformed coefficients, enabling complexity reduction without sacrificing transform quality
2Device complexity
If a sparse transform kernel is applied directly to residual blocks, then computational load is reduced, but transform performance deteriorates
Solution Approach 1:
The first transform kernel is applied as a preliminary action before the sparse transform kernel. This preliminary transformation prepares the data in a form that is more suitable for the subsequent sparse transform, ensuring that important signal characteristics are preserved before the complexity-reducing sparse operation is performed
Solution Approach 2:
The system changes the parameters of the transform process by using two different transform kernels with different characteristics rather than one kernel. The first kernel handles the initial transformation with full complexity, while the second kernel operates sparsely, creating an optimized parameter configuration that balances performance and complexity
3Manufacturing precision
If high-resolution video encoding is performed with standard transform methods, then picture quality is maintained, but resource consumption increases
Solution Approach 1:
The encoding process is segmented to apply different transform strategies to different stages of residual processing. This segmentation enables the system to maintain high picture quality through accurate transformation while reducing resource consumption by using computationally efficient sparse kernels in the second stage
Solution Approach 2:
The transform parameters are changed from a single conventional kernel to a two-kernel sequence with different computational characteristics. This parameter change allows the system to optimize resource consumption by utilizing the sparse structure of the second kernel while maintaining quality through the complementary first transform
Data Source
AI summary
An image decoding method based on a low-complexity transformation is provided. The method comprises: executing a transformation with respect to a residual block based on a first transformation kernel, and executing a transformation with respect to the transformed residual block based on a second transformation kernel having low-complexity, wherein the first transformation kernel is derived based on a prediction block or an encoding unit block.