Image Encoding Transformation Coefficient Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current image compression methods, such as those used in MPEG-4 and H.264/AVC, face inefficiencies in encoding high-resolution video content due to limitations in predicting transformation coefficients in the frequency domain, particularly when neighboring blocks have different coefficients.

Innovation Solution

A method and apparatus for improving image compression efficiency by predicting transformation coefficients in the frequency domain, where only the difference between the predicted and actual coefficients is encoded, utilizing a hierarchical coding structure and various prediction modes to optimize encoding and decoding processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If transformation coefficients are encoded directly without prediction, then encoding is simple, but compression efficiency is poor

Engineering Contradiction:
Improvecompression efficiencyVSAvoidencoding complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by performing prediction of transformation coefficients before encoding. The encoder predicts transformation coefficients of the current block using coefficients from neighboring blocks, and only encodes the difference between predicted and actual coefficients. This preliminary prediction step reduces the amount of data needing encoding, improving compression efficiency while managing complexity through systematic prediction modes.

Inventive Principle:
Principle #10Preliminary action

2Productivity

If prediction is applied when neighboring blocks have similar coefficients, then compression efficiency improves, but prediction fails when blocks have different coefficients

Engineering Contradiction:
Improvecompression efficiencyVSAvoidprediction accuracy
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent applies dynamics by making the prediction process adaptive based on image characteristics. The encoder determines prediction modes (vertical, horizontal, DC, or skip mode) based on the actual transformation coefficients of neighboring blocks. When neighboring blocks have similar coefficients, prediction modes are applied; when they differ significantly, the skip mode is used. This dynamic adaptation ensures prediction accuracy varies with local image content, improving overall compression efficiency while maintaining reliability.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP2451158B1Method of decoding encoded an image and apparatus for encoding an image
Publication Date: 2020.02.26 SAMSUNG ELECTRONICS CO LTD
  • EP2451158B1 patent drawingFigure 1~2
  • EP2451158B1 patent drawingFigure 3
  • EP2451158B1 patent drawingFigure 4

AI summary

Methods and apparatuses for encoding and decoding an image. A transformation coefficient having a low frequency component to be predicted from among transformation coefficients of a transformation block is replaced with a predetermined value, an incomplete residual block is generated by inversely transforming the transformation block including the replaced transformation coefficient, and the transformation coefficient having the low frequency component is predicted by using values of the incomplete residual block.