Research And Practice Of Watermarking Schema Based On Multi-Objective Optimization

With the rapid development of Internet technology and the widespread use of computers,printer and scanners, brings great convenience to people at the same time, also brings theproblem of copyright protection. Digital watermarking technology to protect the copyright ofmultimedia products provides an effective solution, causing more and more attention byresearchers.Classical QIM (Quantization Index Modulation) watermarking algorithm has strongrobustness and easy to realize the advantages, in recent years has been widely studied andused. This algorithm is based on the quantization index modulation algorithm, the paperanalyzes the algorithm is sensitive to the amplitude scaling; at the same time, the embeddingposition and quantization step is hard to decide. So the paper combines genetic algorithm andmulti-objective optimization algorithms to improve the performance for QIM scheme,focusing on the implementation process of research and experimental realization throughMATLAB.This paper summarizes the basic knowledge of digital watermarking technology,introduces the genetic optimization algorithm and multi-objective optimization algorithm,analyze the characteristics of the classic QIM algorithm. Watermarking system referenced inthe genetic algorithm to embed the location and the quantization step as the optimizationparameters, watermark transparency and robustness of the weighting as a performancefunction, the use of genetic algorithms excellent global search capability to find the bestembedding parameters; But also the quantization step size of the algorithm to make theappropriate changes, to improve the classical QIM algorithm against amplitude scaling ofimage.Since transparency and robustness are the watermarking system to optimize theperformance index, which is a multi-objective optimization problem, for addingmulti-objective optimization genetic algorithm, the program is implemented to simplify thedesign of the fitness function and avoid the artificial set goal weight deficiencies. The paperalso focuses on the process of MATLAB-based implementation, and experimental data on theperformance of watermarking schemes for comparison.