Enhancement of human vision sensitivity features for watermarking performance using wavelet packet transformation

• Main Author: Mohammed Shaamala, Abduljabbar Hasan
• Format: Thesis
• Language: English
• Published: 2014
• Subjects: QA76 Computer software
• Online Access: http://eprints.utm.my/id/eprint/48679/1/AbduljabbarHasanMohammedShaamalaPFC2014.pdf; http://eprints.utm.my/id/eprint/48679/; http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:77646?queryType=vitalDismax&query=Enhancement+of+human+vision+sensitivity+features+for+watermarking+performance+using+wavelet+packet+transformation&public=true
• Institution: Universiti Teknologi Malaysia

Digital watermarking has grown extensively in the past few years. It embeds an invisible payload into digital content for the purpose of copyright protection, content authentication, forensic tracking and others. In some applications such as medical, military and law enforcement, even the imperceptible distortion introduced in the watermarking process is unacceptable. The important requirements are to make sure the embedding watermark is undetectable by the human eye, robust against common attack and capacity required for application. Most studies in the field of watermarking based on Human Vision Sensitivity (HVS) features only focus on embedding data into approximate coefficient and have not covered the best combination of the features. The purpose of this research is to determine the optimum region for embedding the watermark, and to enhance watermarking using HVS features. The enhancement approach of sensitivity watermarking is proposed in the process of embedding the watermarks while taking into consideration on maintaining the watermark from distortion. Before the embedding process, the host image is transformed using wavelet packet. The HVS features will try to identify the embedding region inside the coefficient block. In addition, this technique was based on testing 10 different percentages of coefficient regions. This approach exploits vision sensitivity features for embedding high rate payload data into a host image without distortion. Furthermore, the elimination of HVS features was tested to select the best combination of features in order to perform the watermark embedding. The results of this research show the highest embedding rate which is at 3.74 bpp with high imperceptibility rate at 37.1dB compared to others available schemes. The proposed scheme protects the watermark from destruction after attacks or a JPEG compression. It is also discovered that the middle percentages achieved high in capacity, robustness and visual quality. The HVS features also have a significant impact of increasing the performance of watermarking requirements especially on the capacity of embedded message bits.