Encoding device, decoding device, encoding method, and decoding method

The encoding device improves video coding efficiency by deriving and selecting optimal reference images through reconstructed and generated images, utilizing neural networks for transformations, thereby enhancing image quality and processing speed while reducing circuit size.

WO2026155103A1PCT designated stage Publication Date: 2026-07-23PANASONIC INTELLECTUAL PROPERTY CORP OF AMERICA
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
PANASONIC INTELLECTUAL PROPERTY CORP OF AMERICA
Filing Date
2026-01-13
Publication Date
2026-07-23

AI Technical Summary

Technical Problem

Existing video coding technologies face challenges in improving encoding efficiency, image quality, processing load, and circuit size, particularly in selecting appropriate elements or operations such as filters, blocks, motion vectors, and reference pictures.

Method used

The proposed solution involves an encoding device that uses a circuit and memory to derive reconstructed and generated reference images, selecting a collated image from these to derive candidate motion vectors, potentially utilizing neural networks for generating reference images through transformations like scaling, shifting, rotating, or mirroring.

Benefits of technology

This approach enhances encoding efficiency by reducing errors and delays, improving image quality, and optimizing processing speed while minimizing circuit size, thus addressing the limitations of existing technologies.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure JP2026000691_23072026_PF_FP_ABST
    Figure JP2026000691_23072026_PF_FP_ABST
Patent Text Reader

Abstract

An encoding device (100) comprises a circuit (a1) and a memory (a2) connected to the circuit (a1). The circuit (a1): uses the memory (a2) to derive one or more reconstructed reference images, each of which is a reference image derived by reconstructing an image (S3001); generates one or more generated reference images, each of which is a reference image generated from another reference image (S3002); and selects, from among the one or more reconstructed reference images of a plurality of reference images that include the one or more reconstructed reference images and the one or more generated reference images, a collocated image to be used for deriving a motion vector candidate for encoding an image (S3003).
Need to check novelty before this filing date? Find Prior Art