Video Decoder Prediction Buffering for Faster Intra-Inter Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The decoding speed of digital video is reduced due to the division of video data into coding units of various sizes, as intra prediction relies on reconstructed pixel data of reference coding units, leading to inefficiencies in inter prediction control.

Innovation Solution

A decoding device that stores inter prediction data of inter coding units and uses this data to enhance decoding speed by performing inter prediction on each coding unit independently, while also reconstructing intra coding units based on intra prediction data stored in memory.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If intra prediction uses reconstructed pixel data of reference coding units, then prediction accuracy is improved, but decoding speed is reduced

Engineering Contradiction:
Improveprediction accuracyVSAvoiddecoding speed
Core Design Contradiction:
Measurement precisionVSSpeed

Solution Approach 1:

The patent divides the prediction process into separate inter prediction and intra prediction paths. Inter prediction data is generated independently and stored in memory, while intra prediction can proceed separately using its own reference data, eliminating the sequential dependency that slowed down decoding while maintaining prediction accuracy through proper segmentation of prediction units and processes

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs inter prediction in advance and stores the results in memory before intra prediction needs them. This preliminary generation and storage of inter prediction data allows intra prediction to proceed without waiting for inter prediction completion, improving decoding speed while ensuring prediction accuracy through pre-computed reference data

Inventive Principle:
Principle #10Preliminary action

2Loss of substance

If video data is divided into coding units of various sizes, then compression efficiency is improved, but decoding complexity increases

Engineering Contradiction:
Improvecompression efficiencyVSAvoiddecoding complexity
Core Design Contradiction:
Loss of substanceVSDevice complexity

Solution Approach 1:

The patent segments video data into large coding units that are further divided into smaller coding units for prediction processing. By organizing the decoding structure around large coding units with systematic subdivision, the patent manages complexity through hierarchical segmentation while maintaining compression efficiency through adaptive coding unit sizes

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent creates a universal decoding structure where large coding units serve multiple functions: as organizational containers for data management, as prediction processing units, and as memory storage units. This multi-functionality reduces overall system complexity by using a unified structure across different processing stages rather than separate structures for each function

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

3Quantity of substance

If inter prediction data is generated sequentially for each coding unit, then memory usage is reduced, but decoding speed is reduced

Engineering Contradiction:
Improvememory usageVSAvoiddecoding speed
Core Design Contradiction:
Quantity of substanceVSSpeed

Solution Approach 1:

The patent performs inter prediction in advance for multiple coding units and stores results in memory before they are needed for intra prediction or reconstruction. This preliminary action allows parallel processing of multiple coding units simultaneously, improving decoding speed while managing memory through organized pre-computation and storage of prediction data

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS12549707B2Decoding device and operating method thereof
Publication Date: 2026.02.10 POSTECH ACADEMY INDUSTRY FOUNDATION
  • US12549707B2 patent drawing
  • US12549707B2 patent drawing
  • US12549707B2 patent drawing

AI summary

A decoding device includes: an entropy decoder configured to decode a bitstream and generate prediction information and sequence information of each of the plurality of coding units; an inter predictor configured to perform inter prediction on each of inter coding units among the plurality of coding units based on the prediction information; an intra predictor configured to perform intra prediction on each of intra coding units among the plurality of coding units, based on the prediction information, and generate intra prediction data of each of the intra coding units; a memory for storing the inter prediction data of each of the inter coding units; and a reconstructor configured to reconstruct each of the inter coding units, based on the inter prediction data stored in the memory, and reconstruct each of the intra coding units based on the intra prediction data.