Image Embedding via Lattice Division and Asymmetric Assignment

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing information embedding techniques face challenges in embedding and extracting information from images with varying optical densities, particularly in document images containing characters, photographs, and figures, where white regions restrict the applicability, leading to decreased extraction accuracy.

Innovation Solution

An information processing device that divides images into a lattice formation, assigns partial information items uniformly across division images in predetermined directions, and uses discrete cosine transform for embedding, ensuring all partial information items appear equally and can be extracted accurately even in restricted image regions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If information embedding is performed repeatedly in image regions, then extraction accuracy is improved, but the appearance ratio of partial information items decreases when repeatability cycle and applicability cycle are in agreement

Engineering Contradiction:
Improveextraction accuracyVSAvoidappearance ratio of partial information items
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent introduces an offset value to create asymmetric assignment between the repeatability cycle and applicability cycle. By adding this offset, the cycles are deliberately misaligned so they do not coincide, ensuring that partial information items maintain their appearance ratio while still achieving high extraction accuracy through repeated embedding.

Inventive Principle:
Principle #4Asymmetry

2Ease of manufacture

If information is embedded in image regions with large optical density, then embedding is easier, but information cannot be embedded in image regions with small optical density

Engineering Contradiction:
Improveembedding easeVSAvoidapplicability to image regions
Core Design Contradiction:
Ease of manufactureVSAdaptability or versatility

Solution Approach 1:

The patent creates a universal information embedding method that works across different image regions regardless of optical density. By using the offset value to distribute partial information items across the entire image area, the system can embed information in both high-density and low-density regions, making the embedding process universally applicable to all image types including documents with white spaces.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Productivity

If partial information items are assigned to white regions repeatedly, then the repeatability cycle decreases, but extraction accuracy of partial information items falls

Engineering Contradiction:
Improverepeatability cycleVSAvoidextraction accuracy of partial information items
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent uses an offset value to create asymmetric distribution of partial information items across image regions. This prevents systematic assignment of certain information items to unsuitable regions like white spaces, ensuring that all partial information items have equal appearance ratios and maintaining high extraction accuracy while achieving efficient repeatability.

Inventive Principle:
Principle #4Asymmetry

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

The solution enables accurate embedding and extraction of information in images with restricted embeddable regions, improving extraction accuracy by ensuring uniform distribution of partial information items across the image, even in areas with low optical density.

Implementation Method 1

uses discrete cosine transform for embedding

Methodology Applied
Scientific EffectDiscrete cosine transform:

Data Source

PatentEP2017782B1Information processing device, information embedding method, and program
Publication Date: 2018.10.31 RICOH CO LTD
  • EP2017782B1 patent drawingFigure 1
  • EP2017782B1 patent drawingFigure 2
  • EP2017782B1 patent drawingFigure 3A~3B

AI summary

In an information processing device which embeds a number of partial information items produced from predetermined information in an image, an image dividing unit divides an image into a number of division images in a lattice formation to generate the division images. An information assigning unit assigns the partial information items to the division images generated by the image dividing unit. An information embedding unit embeds the partial information items assigned by the information assigning unit, in the division images. The information assigning unit assigns all the partial information items to a set of division images located in a predetermined direction.