Inverse Transform Calculation Using Zero Coefficient Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fast algorithms for inverse transforms in video decompression fail to take advantage of the zero coefficients resulting from quantization, leading to inefficient computation and increased power consumption in devices like handheld electronic devices.
Innovation Solution
The method involves determining and selecting representative computation modes based on the non-zero coefficients of the input data to optimize the inverse transform process, reducing the number of computations required.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If generic fast algorithms are used for inverse transform, then computation time is reduced compared to brute-force methods, but the algorithms fail to take advantage of zero coefficients and cannot further reduce computations
Solution Approach 1:
The patent applies dynamics by making the inverse transform algorithm adaptive rather than static. The system dynamically selects between different computation modes (full computation mode and simplified computation mode) based on the actual characteristics of the input coefficients. This dynamic adaptation allows the algorithm to optimize computation based on real-time data properties, resolving the contradiction between computation speed and algorithm complexity.
Solution Approach 2:
The patent changes the parameter of computation mode based on the input data characteristics. When coefficients are predominantly zero, the system switches to simplified computation mode with fewer operations. When coefficients are dense, it uses full computation mode. This parameter change allows the system to achieve faster computation when possible while maintaining accuracy when necessary, resolving the contradiction between speed and complexity.
2Productivity
If all coefficients are assumed to be non-zero in fast algorithms, then the algorithms are simpler and more generic, but computation cannot be further reduced despite many coefficients being zero after quantization
Solution Approach 1:
The patent applies local quality by treating different regions of the coefficient data differently. Instead of applying a uniform computation approach to all coefficients, the system identifies regions with zero coefficients and applies simplified computations only to those regions, while using full computation for regions with non-zero coefficients. This localized adaptation improves computation efficiency without sacrificing accuracy in regions that require it.
Solution Approach 2:
The patent uses partial action by applying simplified computation modes only when the input data characteristics justify it (i.e., when there are many zero coefficients). When coefficients are dense, the system reverts to full computation mode. This partial application of simplification strategies allows the system to gain efficiency where possible while maintaining accuracy where necessary, resolving the contradiction between computation efficiency and accuracy.
3Use of energy by stationary object
If computation is reduced by exploiting zero coefficients, then power consumption decreases and devices can run at lower frequencies, but the algorithm becomes more complex requiring mode determination and selection
Solution Approach 1:
The patent applies preliminary action by performing a preliminary analysis of the input coefficients to determine their characteristics (density of non-zero coefficients) before committing to a computation mode. This preliminary step allows the system to choose the most energy-efficient computation path in advance, ensuring that power consumption is minimized without requiring complex runtime decisions during the actual transform computation.
Data Source
AI summary
Inverse transforms used in video and image compression/decompression, such as DCT/IDCT used in MPEG-2 and MPEG-4, or the integer transforms used in H.264, are usually calculated with fast algorithms, which only take advantage of the symmetry existing in the transform matrix but ignores the peculiarities in the input data. While these kind of fast algorithms can apply to both forward and inverse transforms, they tend to be inefficient in calculating inverse transforms. In inverse transforms, most of the coefficients become zero after quantization, this invention takes advantage of this fact to further simplify the general fast algorithm and speed up the calculation.