Interlaced Motion Vector Prediction With Four-Field Macroblocks

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video compression and decompression techniques, particularly in interlaced video frames, face limitations such as skewed motion vector predictors and restrictive spatial adaptivity due to median prediction methods and macroblock pairing, which affect the efficiency of motion estimation and compensation.

Innovation Solution

The implementation of techniques that allow for the use of four field motion vectors in interlaced frame-coded pictures, enabling the calculation of motion vector predictors based on valid candidate vectors and considering field polarities, to improve motion vector prediction and adaptivity in interlaced P-frames.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If median prediction method is used for motion vector prediction, then prediction accuracy is improved, but spatial adaptivity is reduced due to macroblock pairing constraints

Engineering Contradiction:
Improvemotion vector prediction accuracyVSAvoidspatial adaptivity
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent divides the macroblock into four separate field motion vectors instead of treating it as a single unit. This segmentation allows each field to have its own motion vector prediction, enabling independent optimization for top and bottom fields while maintaining spatial adaptivity without being constrained by macroblock pairing.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different prediction strategies to different fields within the same macroblock. By allowing separate field motion vectors, the system can optimize prediction accuracy locally for each field based on its specific motion characteristics, rather than applying a uniform prediction approach to the entire macroblock.

Inventive Principle:
Principle #3Local quality

2Adaptability or versatility

If four field motion vectors are used in interlaced frame-coded pictures, then spatial adaptivity and prediction accuracy are improved, but computational complexity increases

Engineering Contradiction:
Improvespatial adaptivityVSAvoidcomputational complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent performs preliminary classification to determine whether a macroblock should be processed with four field motion vectors or fewer. By pre-determining the appropriate motion vector mode based on motion characteristics, the system avoids unnecessary computational overhead for blocks that don't require full four-field processing, thus managing complexity while maintaining adaptivity where needed.

Inventive Principle:
Principle #10Preliminary action

3Ease of manufacture

If motion vector predictors are calculated using traditional methods, then encoding simplicity is maintained, but prediction accuracy suffers due to skewed predictors in interlaced frames

Engineering Contradiction:
Improveencoding simplicityVSAvoidmotion vector prediction accuracy
Core Design Contradiction:
Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent introduces dynamic adaptation in motion vector prediction by allowing the prediction mode to change based on the interlaced frame characteristics. The system dynamically selects between different prediction approaches (including four field motion vectors) depending on the motion patterns detected, rather than using a fixed prediction method, thus improving accuracy while maintaining reasonable encoding complexity.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS7924920B2Motion vector coding and decoding in interlaced frame coded pictures
Publication Date: 2011.04.12 MICROSOFT TECHNOLOGY LICENSING LLC
  • US7924920B2 patent drawing
  • US7924920B2 patent drawing
  • US7924920B2 patent drawing

AI summary

In one aspect, an encoder/decoder receives information for four field motion vectors for a macroblock in an interlaced frame-coded, forward-predicted picture and processes the macroblock using the four field motion vectors. In another aspect, an encoder/decoder determines a number of valid candidate motion vectors and calculates a field motion vector predictor. The encoder/decoder does not perform a median operation on the valid candidates if there are less than three of them. In another aspect, an encoder/decoder determines valid candidates, determines field polarities for the valid candidates, and calculates a motion vector predictor based on the field polarities. In another aspect, an encoder/decoder determines one or more valid candidates, determines a field polarity for each individual valid candidate, allocates each individual valid candidate to one of two sets (e.g., opposite polarity and same polarity sets) depending on its field polarity, and calculates a motion vector predictor based on the two sets.