Switchable Filter Index for Video Resampling Accuracy

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding techniques face challenges in efficiently compressing video data while maintaining high quality, particularly when dealing with reference picture resampling across different pixel grids, as conventional filters perform poorly when used for fractional accuracy in such scenarios.

Innovation Solution

The implementation of a system that identifies filter index information for a current block and determines if the reference picture size differs from the current picture size, using a default filter index for sample rate conversion processes instead of the current filter index, thereby improving encoding and decoding efficiency by selecting appropriate filters for interpolation and resampling tasks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If conventional filters are used for reference picture resampling, then the filtering process is simple, but the fractional accuracy in motion prediction deteriorates

Engineering Contradiction:
Improvefiltering process complexityVSAvoidfractional accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent changes the filter parameters (coefficients) based on the resampling context. When reference picture resampling is detected, the system switches to a second set of filter coefficients optimized for resampling operations, whereas conventional filters use a fixed first set of coefficients. This parameter adaptation resolves the contradiction by providing high fractional accuracy when needed while maintaining simple filtering when resampling is not required.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The filtering system transitions from a static, fixed-filter approach to a dynamic, context-adaptive approach. The filter dynamically selects between different coefficient sets based on whether reference picture resampling is detected, allowing the system to optimize fractional accuracy for motion prediction when resampling occurs while maintaining simplicity in other scenarios.

Inventive Principle:
Principle #15Dynamics

2Productivity

If different sized reference pictures are used for inter-prediction, then video compression efficiency is improved, but filtering accuracy deteriorates

Engineering Contradiction:
Improvevideo compression efficiencyVSAvoidfiltering accuracy
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The system detects when reference pictures of different sizes are used (indicating resampling) and automatically switches to a second set of filter coefficients designed specifically for resampling operations. This allows the system to maintain high filtering accuracy even when compression efficiency benefits from using differently sized reference pictures.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent introduces an intermediary detection mechanism that identifies when reference picture resampling is occurring. This detection acts as a mediator between the compression efficiency benefit of using different sized pictures and the filtering accuracy requirement, triggering the appropriate filter coefficient selection to maintain accuracy across both scenarios.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Device complexity

If a fixed filter index is used for all blocks, then processing complexity is reduced, but encoding and decoding results deteriorate

Engineering Contradiction:
Improveprocessing complexityVSAvoidencoding and decoding results
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

Instead of using a fixed filter index for all blocks, the system dynamically changes the filter index based on whether reference picture resampling is detected for each block. This allows the system to maintain low processing complexity through simple detection logic while significantly improving encoding and decoding results by applying the appropriate filter coefficients for each block's specific resampling context.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS11563933B2Reference picture resampling with switchable filters
Publication Date: 2023.01.24 QUALCOMM INC
  • US11563933B2 patent drawing
  • US11563933B2 patent drawing
  • US11563933B2 patent drawing

AI summary

Techniques are described for video encoding and decoding using reference picture resampling with switchable filters. One example involves obtaining a current picture and a reference picture, identifying filter index information for a current block of the current picture, and determining that a first picture size value of the current picture and a second picture size value of the reference picture are different. Based on the determining that the first picture size value of the current picture and the second picture size value of the reference picture are different, performing a resampling process using a default filter index in place of a current filter index identified by the filter index information. Additional examples can use the current filter index identified by the filter index information in subsequent blocks. In various examples, the current filter index can be derived or signaled.