8-Point IDCT Hardware Unit Even Odd Factorization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional 8-point discrete cosine transform (DCT) methods in video compression do not achieve optimal coding gain due to limitations in factorization and implementation complexity, particularly in power-sensitive devices like mobile devices.
Innovation Solution
The implementation of an 8-point DCT with specific relationships between internal and scaled factors, allowing for orthogonal or near-orthogonal transformations that reduce implementation complexity and enhance coding efficiency by factorizing internal factors and incorporating external scaled factors in quantization processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional 8-point DCT methods are used, then implementation is simpler, but coding gain is not optimal
Solution Approach 1:
The 8-point DCT is segmented into even and odd portions, each processed separately with dedicated factorizations. The even portion uses internal factors (A, B) with scaled factor μ, while the odd portion uses internal factors (G, D, E, Z) with scaled factor η. This segmentation allows optimal factorization for each portion, improving coding gain while managing complexity through modular structure.
Solution Approach 2:
The patent changes the parameter relationships by defining specific mathematical relationships between internal factors and scaled factors. The scaled factor μ equals the square root of the sum of squares of internal factors A and B, and similarly for η with factors G, D, E, and Z. These parameter relationships enable orthogonal or near-orthogonal transformations that optimize coding gain.
2Productivity
If orthogonal or near-orthogonal transformations are implemented with specific factor relationships, then coding efficiency is enhanced, but implementation complexity increases
Solution Approach 1:
The patent extracts the complexity of maintaining orthogonality into separate scaled factors (μ and η) that are removed from the core DCT computation. These scaled factors are calculated once based on the internal factors and then applied as simple multiplications, rather than requiring complex real-time orthogonal transformation calculations, thus enhancing coding efficiency while managing implementation complexity.
Solution Approach 2:
The internal factors and scaled factors are pre-calculated and stored in lookup tables before the actual DCT operation. The relationships between factors are established in advance (μ = √(A² + B²), η = √(G² + D²) = √(E² + Z²)), allowing the transform to proceed with simple table lookups and multiplications rather than complex computations during video encoding, improving coding efficiency.
3Device complexity
If factorization with internal and scaled factors is applied, then implementation complexity is reduced, but coding gain may be compromised
Solution Approach 1:
The patent merges the factorization approach with orthogonal transformation requirements by carefully designing the relationships between internal factors and scaled factors. The even portion combines factors A and B with scaled factor μ, while the odd portion combines factors G, D, E, and Z with scaled factor η. This merging ensures that the factorized implementation maintains orthogonal properties, preserving coding gain while reducing implementation complexity.
Data Source
AI summary
In general, techniques are described for implementing an 8-point inverse discrete cosine transform (IDCT). An apparatus comprising an 8-point inverse discrete cosine transform (IDCT) hardware unit may implement these techniques to transform media data from a frequency domain to a spatial domain. The 8-point IDCT hardware unit includes an even portion comprising factors A, B that are related to a first scaled factor (μ) in accordance with a first relationship. The 8-point IDCT hardware unit also includes an odd portion comprising third, fourth, fifth and sixth internal factors (G, D, E, Z) that are related to a second scaled factor (η) in accordance with a second relationship. The first relationship relates the first scaled factor to the first and second internal factors. The second relationship relates the second scaled factor to the third, fourth, fifth and sixth internal factors.