Video Decoder DPB Memory Management via Chroma Format and Bit Depth

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding technologies face challenges in efficiently managing memory in the decoded picture buffer (DPB) during intra random access point (IRAP) decoding, particularly due to changes in color format and bit depth, which can lead to buffer overflow and decoding performance issues.

Innovation Solution

The proposed solution involves determining the value of the NoOutputOfPriorPicsFlag based on changes in the chroma format and bit depth of pictures, allowing for accurate anticipation and management of memory needs in the DPB, thereby controlling the output and removal of pictures without affecting decoder performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If pictures are retained in the decoded picture buffer (DPB) for reference, then decoding accuracy is improved, but memory consumption increases and buffer overflow may occur

Engineering Contradiction:
Improvedecoding accuracyVSAvoidmemory consumption
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The patent changes the parameter used to determine picture retention from spatial resolution to chroma format and bit depth. When chroma format or bit depth changes between current and reference pictures, the NoOutputOfPriorPicsFlag is set to flush the DPB, preventing buffer overflow while maintaining decoding accuracy for pictures with matching characteristics.

Inventive Principle:
Principle #35Parameter changes

2Quantity of substance

If the NoOutputOfPriorPicsFlag is set to flush DPB, then buffer overflow is prevented, but decoding performance may be affected due to loss of reference pictures

Engineering Contradiction:
Improvebuffer capacityVSAvoiddecoding performance
Core Design Contradiction:
Quantity of substanceVSProductivity

Solution Approach 1:

The patent applies local quality by selectively flushing the DPB based on specific picture characteristics (chroma format and bit depth) rather than universally. Only pictures with different chroma format or bit depth are flushed, while pictures with matching characteristics are retained as references, thus maintaining decoding performance where possible.

Inventive Principle:
Principle #3Local quality

3Device complexity

If spatial resolution is used to determine NoOutputOfPriorPicsFlag, then memory management is simplified, but accuracy in anticipating memory needs decreases due to color format and bit depth changes

Engineering Contradiction:
Improvememory management complexityVSAvoidmemory need prediction accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent changes the determining parameters from spatial resolution to chroma format and bit depth. This provides more accurate prediction of memory needs because chroma format and bit depth directly affect the actual memory consumption of picture data, even when spatial resolution remains constant.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3090557B1Improved inference of nooutputofpriorpicsflag in video coding
Publication Date: 2019.12.18 QUALCOMM INC
  • EP3090557B1 patent drawingFigure 1A
  • EP3090557B1 patent drawingFigure 1B
  • EP3090557B1 patent drawingFigure 2A

AI summary

An apparatus for coding video information according to certain aspects includes a processor configured to determine a value of a flag associated with a current picture of a current layer to be decoded, the flag indicating whether pictures in a decoded picture buffer (DPB) should be output, wherein the current picture is an intra random access point (IRAP) picture that starts a new coded video sequence (CVS) and wherein the determination of the value of the flag is based on at least one of: (1) the chroma format of the current picture and the chroma format of the preceding picture, (2) the bit depth of the luma samples of the current picture and the bit depth of the luma samples of the preceding picture, or (3) the bit depth of the chroma samples of the current picture and the bit depth of the chroma samples of the preceding picture.