Inverse Discrete Cosine Transform System Reducing Summation Operations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital video (DV) codecs require extensive hardware and power to perform inverse discrete cosine transform (IDCT) operations, leading to increased costs and prolonged decompression times due to the need for numerous summations and separate hardware for 8-8 and 2-4-8 IDCT modes.
Innovation Solution
The proposed IDCT system reduces calculations by exploiting symmetry characteristics to sum partial output matrices, using a 2N×2N IDCT procedure that divides the output matrix into sub-matrices with symmetry relations, thereby reducing the number of summation devices and operations, and integrating 8-8 and 2-4-8 IDCT processes into a single hardware platform.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional IDCT procedure is used to transform DCT coefficients back to output matrix elements, then accurate decompression is achieved, but hardware cost and power consumption increase due to numerous summation operations
Solution Approach 1:
The patent divides the 8×8 IDCT computation into multiple 4×4 sub-matrix computations. Each 4×4 sub-matrix is processed independently using segmented summation operations, reducing the complexity of individual computational blocks while maintaining overall accuracy through systematic combination of sub-results.
Solution Approach 2:
The patent performs partial summation operations by computing only necessary intermediate sums for each 4×4 sub-matrix rather than performing all 4096 summations in the conventional approach. This partial action reduces computational load while achieving sufficient precision for video decompression.
2Reliability
If separate hardware is used for 8-8 and 2-4-8 IDCT modes, then both modes can be processed accurately, but hardware cost and production complexity increase
Solution Approach 1:
The patent designs a unified IDCT processing architecture that can handle both 8-8 and 2-4-8 modes through the same hardware platform. The segmented 4×4 sub-matrix computation approach provides a universal framework that adapts to different DCT modes without requiring separate dedicated hardware for each mode.
Solution Approach 2:
The patent implements dynamic mode switching capability where the same hardware can adapt its operation based on the input mode (8-8 or 2-4-8). The computational pipeline dynamically adjusts its processing based on the detected DCT mode, allowing one hardware system to perform multiple functions.
3Manufacturing precision
If conventional IDCT computation with 4096 summation operations is performed, then complete output matrix is obtained, but decompression time is prolonged
Solution Approach 1:
The patent segments the computationally intensive IDCT operation into parallel 4×4 sub-matrix processing tasks. By dividing the 4096 summation operations into smaller, independent segments that can be processed concurrently, the overall decompression time is reduced while maintaining complete output matrix generation.
Solution Approach 2:
The patent computes partial output matrices through selective summation operations rather than performing all possible summations. This partial computation approach achieves sufficient output completeness for video reconstruction while significantly reducing the total number of operations required.
Data Source
AI summary
A data processing system transforms an input matrix with transform coefficients through an inverse discrete cosine transform procedure, and it consequentially obtains an output matrix. The input matrix comprises 2N×2N of discrete cosine transform coefficients, and the output matrix comprises 2N×2N of output data; both are sequentially indexed by the row and column index. The output matrix is obtained by summing up a plurality of partial output matrixes; each partial output matrix comprises a first partial sub-output matrix and at least one other partial sub-output matrix. The first partial sub-output matrix comprises a plurality of partial output data that has a predetermined symmetry characteristic. The partial output matrixes possess a number of symmetry relations. The first and other partial sub-output matrixes of each partial output matrix have one symmetry relation. The data processing system comprises a system controller, a calculating unit, a symmetry duplicator, a summation device and four summed partial sub-output matrix register.