Image Data Information Embedding via Frequency Domain Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for embedding information in image data face challenges, particularly in images where information cannot be embedded due to their characteristics, leading to inefficiencies and potential image degradation.
Innovation Solution
An information embedding apparatus that extracts feature quantities from image data, calculates codes based on these features, and modifies the image data to embed the codes efficiently, using a combination of code extraction, embedding, and decoding units to ensure compatibility and minimize image alteration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If information is embedded in image data by changing gray values, then information can be hidden in the image, but information cannot be embedded in some images due to their characteristics
Solution Approach 1:
The patent transforms image data into the frequency domain using Fourier transform, changing the representation parameters from spatial domain gray values to frequency domain coefficients. This parameter transformation enables information embedding in the frequency domain, which is applicable to various image characteristics including those where spatial domain embedding fails. The feature quantities are selected from frequency domain coefficients, and information is embedded by modifying these coefficients rather than direct gray values.
2Productivity
If conventional embedding methods are used, then embedding process is simple, but information cannot be embedded in all image types
Solution Approach 1:
The patent introduces Fourier transform as an intermediary transformation process between the original image data and the embedding target. This intermediary step converts spatial domain data to frequency domain data, creating a new representation space where information embedding can be performed on various image types. The feature quantity extraction and code embedding operations are performed in this intermediate frequency domain, enabling universal applicability while maintaining a systematic process.
3Loss of information
If gray values are changed to embed information, then information is embedded, but image quality may be degraded
Solution Approach 1:
The patent applies different treatment to different frequency components of the image data. Feature quantities are selectively extracted from specific frequency domain coefficients, and only these selected coefficients are modified for information embedding. This local quality approach ensures that important frequency components (which contribute more to image quality) are preserved, while less critical components are used for embedding, thereby maintaining image quality while achieving information embedding.
Data Source
AI summary
In an information embedding apparatus, an image selecting unit selects original image data from an original image data group, a decoder detects a code from the selected original image data, and a candidate code determining unit calculates a candidate code. A candidate code embedding unit embeds the candidate code in the selected original image data.