Evaluating Performance of Text Watermarking in Digital Images

Abstract

Copyright protection and ownership proof of digital multimedia (such as text, audio, images and videos) nowadays are achieved using digital watermarking technique. This thesis, used a text watermarking method for protecting the property rights and ownership judgment of multimedia. However, it focuses on the process of hiding a small text or owner information in images. In this thesis, two watermarking domains techniques are introduced which are spatial domain and frequency domain by comparing three different watermarking techniques which are LSB, DCT and FFT. The LSB is considered the basic and traditional technique which is a special domain watermarking technique, while DCT and FFT are the frequency domain watermarking technique. The Evaluation extended to include investigation of different types of watermarking attacks namely geometrical attacks and noise attacks. Also, different types of image formats are evaluated which are the BMP, JPG, PNG, and TIFF. Finally watermark extraction results and image evaluation performance metrics (PSNR, NCC and MSSIM) are mathematically described, and the impact of watermarking techniques, types of image format, and different types of attacks on these evaluation performance metrics. The used text watermarking method is suitable for SMS text messages embedded in digital image for the purpose of copyright protection and proprietary rule applications, rather than applying steganography that is used for secure information exchange purposes.