Transcoder Motion Vector Reuse and Frequency Domain Downsampling

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video transcoding methods from MPEG-2 to MPEG-4 format are computationally intensive due to the need for motion vector estimation and spatial resolution reduction, particularly in the context of 3G wireless infrastructure and video streaming applications, where high bitrate MPEG-2 content needs to be converted to lower bitrate, progressive standards like MPEG-4.

Innovation Solution

The proposed solution involves motion vector reuse by best predictor selection, motion vector refinement through search window adaptation, frequency domain downsampling with spatial averaging of frame-DCT blocks and field-DCT blocks, and a mixture of one-dimensional de-interlacing IDCT with IDCT plus downsampling to minimize computational complexity and drift artifacts.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If motion vector estimation is performed in the transcoder, then motion compensation accuracy is improved, but computational complexity increases significantly

Engineering Contradiction:
Improvemotion compensation accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by reusing motion vectors from the MPEG-2 decoder as initial estimates for MPEG-4 motion compensation. Instead of performing exhaustive motion vector estimation from scratch, the system prepares and utilizes previously computed motion vectors, significantly reducing the computational burden while maintaining acceptable compensation accuracy.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent changes the parameter of motion vector precision by using integer-pixel accuracy from MPEG-2 motion vectors rather than attempting to achieve sub-pixel accuracy through exhaustive search. This parameter change reduces computational complexity while accepting a trade-off in compensation precision that is acceptable for the application.

Inventive Principle:
Principle #35Parameter changes

2Quantity of substance

If spatial down-sampling is performed to reduce resolution, then output bitrate is reduced, but drift artifacts increase

Engineering Contradiction:
ImprovebitrateVSAvoiddrift artifacts
Core Design Contradiction:
Quantity of substanceVSObject-generated harmful factors

Solution Approach 1:

The patent introduces an intermediary approach by using motion-compensated temporal filtering as a mediator between the high-bitrate MPEG-2 source and low-bitrate MPEG-4 output. This intermediary process reduces temporal redundancy and minimizes drift artifacts during the bitrate reduction process, acting as a buffer that smooths the transition between different bitrates.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent applies segmentation by processing video in terms of macroblocks and utilizing field-based temporal filtering. By segmenting the video into manageable units and applying targeted filtering operations, the system reduces drift artifacts while maintaining efficient bitrate reduction.

Inventive Principle:
Principle #1Segmentation

3Adaptability or versatility

If conventional transcoding methods are used, then format conversion is achieved, but processing time increases

Engineering Contradiction:
Improveformat conversion capabilityVSAvoidprocessing time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent applies universality by designing a transcoder that performs multiple functions: it decodes MPEG-2, performs motion compensation for MPEG-4, conducts spatial down-sampling, and applies temporal filtering all within a unified architecture. This multi-functional approach eliminates the need for separate processing stages and reduces overall processing time while maintaining format conversion capability.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent merges several processing operations into a single integrated system. The MPEG-2 decoder, motion compensation unit, down-sampling filter, and temporal filtering operations are combined into one cohesive transcoding pipeline, allowing simultaneous execution of multiple tasks and significantly reducing processing time compared to sequential conventional methods.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentUS7412003B2Transcoders and methods
Publication Date: 2008.08.12 TEXAS INSTRUMENTS INC
  • US7412003B2 patent drawing
  • US7412003B2 patent drawing
  • US7412003B2 patent drawing

AI summary

Transcoding as from MPEG-2 SDTV to MPEG-4 CIF reuses motion vectors and downsamples in the frequency (DCT) domain with differing treatments of frame-DCT and field-DCT blocks, and alternatively uses de-interlacing IDCT with respect to the row dimension plus deferred column downsampling for reference frame blocks.