Divided Picture Encoding Reducing Deblocking Complexity

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In video coding, particularly in H.264, the deblocking filter used to reduce block distortion contributes significantly to processing complexity, and existing methods to reduce this complexity, such as rearranging pixels or pixel groups, either increase bit overhead or decrease prediction accuracy, thereby deteriorating coding efficiency.

Innovation Solution

The method involves generating divided pictures by rearranging pixels or pixel groups and selecting an encoded divided picture with high correlation as a reference picture for inter divided-picture encoding, eliminating the need to encode a reference picture index, thereby reducing computational complexity and bit overhead while maintaining coding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a deblocking filter is applied to the entire frame to suppress block distortion, then the prediction efficiency is improved, but the computational complexity is increased

Engineering Contradiction:

Improveprediction efficiencyVSAvoidcomputational complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The picture is divided into multiple divided pictures by rearranging pixels or pixel groups. Each divided picture is encoded separately using inter-divided-picture predictive encoding, which reduces block distortion locally without requiring a global deblocking filter, thereby reducing computational complexity while maintaining prediction efficiency

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent combines intra-divided-picture encoding and inter-divided-picture encoding in a hybrid approach. By using both encoding methods on different divided pictures, the system achieves better prediction efficiency and reduced block distortion without the need for extensive deblocking filtering

Inventive Principle:

Principle #40Composite materials

2Device complexity

If pixel rearrangement is performed to reduce deblocking filter complexity, then the processing complexity is reduced, but the coding efficiency deteriorates due to increased bit overhead

Engineering Contradiction:

Improvedeblocking filter complexityVSAvoidbit overhead

Core Design Contradiction:

Device complexityVSLoss of information

Solution Approach 1:

The divided pictures are designed to be self-contained with reference pictures that can be predicted from previously decoded divided pictures. This self-referential structure eliminates the need for additional bit overhead to transmit reference picture information, as the decoder can reconstruct references using the same pixel rearrangement pattern

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent changes the spatial arrangement parameter by rearranging pixels into divided pictures with specific patterns (e.g., interlaced, block-based). This parameter change enables the use of inter-divided-picture prediction without increasing bit overhead, as the rearrangement pattern itself becomes the organizing principle for reference selection

Inventive Principle:

Principle #35Parameter changes

3Device complexity

If pixel rearrangement is performed to reduce deblocking filter complexity, then the processing complexity is reduced, but the prediction accuracy decreases

Engineering Contradiction:

Improvedeblocking filter complexityVSAvoidprediction accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

Reference pictures are prepared in advance by decoding previous divided pictures using the same pixel rearrangement pattern. This preliminary preparation ensures that when inter-divided-picture prediction is performed, high-quality reference data is already available, maintaining prediction accuracy without requiring complex deblocking operations

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces divided pictures as an intermediary representation between the original picture and the final decoded output. This intermediary form with rearranged pixels serves as both the encoding structure and the reference structure, enabling accurate prediction while avoiding the need for heavy deblocking filtering

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentEP2775718B1Method, device, and program for coding and decoding of images

Publication Date: 2019.07.10 NIPPON TELEGRAPH & TELEPHONE CORP

Figure 1
Figure 2A~2D
Figure 3(A)~3(C)

AI summary

Compared to conventional intra divided-picture encoding, deterioration in the coding efficiency is suppressed and the encoding computational complexity and the decoding computational complexity are reduced. A divided picture generating unit divides an input encoding target picture into blocks of the same size and generates divided pictures having the same size by collecting pixels having the same relative position within each block. An intra divided-picture encoding processing unit performs intra divided-picture encoding on part of the divided pictures. An inter divided-picture encoding processing unit performs inter divided-picture encoding using another encoded divided picture as a reference picture. When there are a plurality of candidates for the reference picture, a correlation direction calculating unit obtains a combination in which a correlation of a pixel on an original picture is high among combinations of encoded divided pictures and their reference pictures. A reference picture selecting unit selects an encoded divided picture in a direction in which a correlation with an encoding target divided picture is high as the reference picture to be used in the inter divided-picture encoding processing unit.