Sequency Wavelet Image Watermarking Robustness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image watermarking techniques are vulnerable to attacks such as JPEG compression, median filtering, and blurring, leading to poor extraction of watermarks, which compromises ownership claims in multimedia contents.
Innovation Solution
A new sequency and wavelet transform-based watermarking technique that uses sequencies of host and watermark images to devise a robust embedding scheme, employing an Adaptive Parabolic Gain adjustment and embedding the watermark in the LL band of the host image's wavelet transform, while maintaining perceptual signal-to-noise ratio and quality, and includes a watermark extraction algorithm that accounts for attack impacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional watermarking techniques are used, then the watermark can be embedded in the image, but the watermark extraction becomes poor after attacks like JPEG compression, median filtering, and blurring
Solution Approach 1:
The patent applies parameter changes by transforming the watermark embedding from spatial domain to wavelet transform domain, specifically using the LL sub-band. The watermark is embedded by modifying wavelet coefficients rather than pixel values, and the gain adjustment parameter is dynamically adapted based on local image characteristics. This transformation to different parameter space (frequency domain) provides robustness against spatial attacks like JPEG compression, median filtering, and blurring while maintaining extraction reliability.
2Measurement precision
If the watermark is embedded strongly to improve extraction accuracy, then the watermark can be extracted more accurately, but the image quality degrades
Solution Approach 1:
The patent implements local quality by using adaptive gain adjustment where the watermark embedding strength varies across different regions of the image. The gain parameter is locally adapted based on the statistical properties of the host image and watermark in each wavelet coefficient block. This allows stronger embedding in regions where it is tolerable and weaker embedding in sensitive regions, thereby maintaining extraction accuracy while minimizing overall image quality degradation and achieving PSNR values above 30 dB.
3Ease of manufacture
If the watermark is embedded in the spatial domain, then the embedding process is simple, but the watermark is vulnerable to attacks and extraction is poor
Solution Approach 1:
The patent replaces the mechanical spatial domain embedding approach with a wavelet transform-based frequency domain approach. Instead of directly modifying pixel values in the spatial domain, the system transforms the image and watermark into the wavelet domain, embeds the watermark in the LL sub-band, and then transforms back. This substitution of the embedding mechanism from spatial to frequency domain provides robustness against attacks while maintaining computational feasibility through efficient wavelet transform algorithms.
4Productivity
If conventional watermarking is used, then the process is fast, but the extracted watermark is not recognizable after attacks
Solution Approach 1:
The patent applies segmentation by dividing the image and watermark into blocks and processing them in the wavelet domain. The wavelet transform naturally segments the image into different frequency sub-bands (LL, LH, HL, HH), and the watermark embedding is performed on the LL sub-band which contains the low-frequency components. This segmentation in the frequency domain allows the watermark to be embedded in a way that is resilient to attacks while maintaining processing efficiency through block-wise operation and leveraging the fast wavelet transform algorithm.
Data Source
AI summary
This invention is a new approach for the image watermarking in the wavelet transform domain based on sequency of the host and watermark image. For each sub-band a first transform level of the host image is thresholded and binarized. Sequencies of thresholded and binarized data host image are compared with sequencies of the discrete wavelet transformed watermark image to form a watermarking sequency mask. The watermarked wavelet domain data is formed by combining data elements of the discrete wavelet transformed host image with corresponding data elements of the wavelet transformed watermark image as filtered by the watermarking mask. A reverse process can extract the watermark with a high degree of accuracy even after attack upon the watermarked host image.