8x8 Integer Transform for Low-Complexity Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital video compression methods, such as those using the discrete cosine transform (DCT), face high computational complexity, particularly in 16-bit operations, which hinders efficient video coding and motion compensation.
Innovation Solution
The implementation of low-complexity 8×8 transformation through bit shifting and round-off techniques, specifically using the AVS method with modified forward transforms, quantization, and inverse quantization processes, reduces computational overhead while maintaining signal-to-noise ratios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If standard 8×8 DCT transformation is used in 16-bit operations, then transformation accuracy is maintained, but computational complexity becomes excessively high
Solution Approach 1:
The patent segments the 8×8 DCT transformation into two independent 4×4 transformations. By dividing the large transform into smaller sub-transforms, the computational complexity is reduced while maintaining the overall transformation accuracy through proper recombination of results.
Solution Approach 2:
The patent extracts and removes unnecessary computational operations from the standard 16-bit DCT process. Specifically, it eliminates redundant multiplication operations and optimizes the transformation steps to reduce complexity while preserving the essential transformation functionality.
2Productivity
If quantization step is increased to reduce bit rate, then compression efficiency improves, but signal quality deteriorates
Solution Approach 1:
The patent implements dynamic quantization where the quantization step is adaptively adjusted based on the frequency component being processed. Different quantization steps are applied to different frequency bands, allowing optimization of both compression efficiency and signal quality through adaptive rather than fixed quantization.
Solution Approach 2:
The patent applies different quantization characteristics to different frequency components within the transform block. High-frequency coefficients receive different quantization treatment compared to low-frequency coefficients, allowing local optimization of quality where needed while achieving overall compression efficiency.
3Measurement precision
If 16-bit operations are used for transformation, then processing precision is maintained, but computational overhead increases significantly
Solution Approach 1:
The patent segments the 16-bit transformation operations into smaller 4×4 sub-transforms that can be processed more efficiently. This segmentation allows the system to maintain processing precision through proper fixed-point arithmetic while significantly reducing the computational overhead associated with full 16-bit operations.
Solution Approach 2:
The patent changes the operational parameters by using fixed-point arithmetic instead of full 16-bit floating-point operations. This parameter change maintains the necessary processing precision for video compression while dramatically reducing computational overhead and processing time.
Data Source
AI summary
Low complexity (16 bit arithmetic) video compression has 8×8 block with transforms using 8×8 integer matrices and quantization with look up table scalar plus constant right shift for all quantization steps. Inverse quantization also a look up table scalar plus right shift dependent upon the quantization step and inverse transform using the 8×8 integer matrices.