ART2-based genetic watermarking

A genetic watermarking approach based on ART2 neural network is proposed in the paper. This approach uses an ART2 neural network to classify 8/spl times/8 DCT blocks of images in training sets. For each cluster, genetic algorithm (GA) is then performed to find out the optimal coefficients for waterm...

Full description

Saved in:
Bibliographic Details
Main Authors: Ying-Lan Chang, Koun-Tem Sun, Yueh-Hong Chen
Format: Conference Proceeding
Language:English
Subjects:
Computer science
Constraint optimization
Discrete cosine transforms
Genetic algorithms
Internet
Neural networks
Optimized production technology
Robustness
Sun
Watermarking
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A genetic watermarking approach based on ART2 neural network is proposed in the paper. This approach uses an ART2 neural network to classify 8/spl times/8 DCT blocks of images in training sets. For each cluster, genetic algorithm (GA) is then performed to find out the optimal coefficients for watermark embedding. All the results are recorded in a table, called optimal position table (OPT). According to the OPT table, the coefficients for watermark embedding can be decided straightforward. Two features of the proposed approach make itself a suitable enhancement for genetic watermarking. First, it has the ability to keep and refine the results obtained from genetic watermarking. Second, the proposed method greatly increases the speed of genetic watermarking so that genetic watermarking can be used in practice. The experimental results shows that the watermarked images are perceptually equal to the originals, and that the watermarks are still detectable after low pass filtering, high pass filtering and JPEG compression.
ISSN:1550-445X
2332-5658
DOI:10.1109/AINA.2005.122