Video Encoder Boundary Pixel Smoothing for Prediction Quality

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video encoding technologies face challenges in generating proper prediction images, particularly when reference regions extend beyond the boundaries of reference pictures, leading to unnatural patterns and increased code amounts.

Innovation Solution

An encoder and decoder system that derives motion vectors, performs padding and smoothing on pixels outside the boundary of reference pictures using pixels inside the boundary, and generates prediction images using these processed pixels, thereby reducing unnatural patterns and improving image quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If padding is performed using pixels inside the boundary to generate pixels outside the boundary, then the reference region can be properly formed, but unnatural patterns occur in the prediction image

Engineering Contradiction:

Improveprediction image qualityVSAvoidunnatural patterns

Core Design Contradiction:

Manufacturing precisionVSObject-generated harmful factors

Solution Approach 1:

The patent applies smoothing filters to the padded pixels outside the boundary, changing the parameter of pixel intensity distribution to reduce unnatural patterns. The smoothing operation modifies the raw padded values to create more natural transitions at the boundary region.

Inventive Principle:

Principle #35Parameter changes

2Adaptability or versatility

If reference regions extend beyond the boundary of reference pictures, then motion compensation can be performed, but the prediction image generation becomes difficult

Engineering Contradiction:

Improvemotion compensation capabilityVSAvoidprediction image generation

Core Design Contradiction:

Adaptability or versatilityVSManufacturing precision

Solution Approach 1:

The patent performs padding to generate pixels outside the boundary before the actual prediction image generation process. This preliminary action ensures that the reference region is complete and ready for motion compensation operations.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent applies smoothing filters to the padded pixels to modify their intensity values, making the extended reference region more compatible with the actual picture content and improving prediction accuracy.

Inventive Principle:

Principle #35Parameter changes

3Manufacturing precision

If smoothing is performed on pixels outside the boundary, then unnatural patterns are reduced, but processing complexity increases

Engineering Contradiction:

Improveprediction image qualityVSAvoidprocessing complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent applies smoothing filters specifically to the padded pixels outside the boundary, rather than processing the entire reference picture. This localized approach reduces the overall processing complexity while still improving the prediction image quality where it is most needed.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS10904576B2Encoder, decoder, encoding method, and decoding method for generating a prediction image using a plurality of pixels generated outside of a boundary of a reference picture on which smoothing has been performed

Publication Date: 2021.01.26 PANASONIC INTELLECTUAL PROPERTY CORP OF AMERICA

AI summary

An encoder is an encoder which encodes a video, and includes: circuitry; and memory. The circuitry, using the memory, derives a motion vector of a current block included in a current picture, by referring to a reference picture different from the current picture; when a reference region indicated by the motion vector includes a region outside a boundary of the reference picture, performs padding for generating a plurality of pixels outside the boundary in the reference region, using one of a plurality of pixels inside the boundary of the reference picture, to generate the plurality of pixels outside the boundary; performs smoothing on the plurality of pixels outside the boundary; and generates a prediction image, using the plurality of pixels outside the boundary on which the smoothing has been performed.