In recent years, services on the Internet have greatly improved and are more reliable than before. However, the easy downloads and duplications on the Internet have created a rush of illicit reproductions. Undoubtedly, the rights of ownership are violated and vulnerable to the predators that stalk the Internet. Therefore, protection against these illegal acts has become a mind-boggling issue. Previously, artists and publishers painstakingly signed or marked their products to prevent illegal use. However with the invention of digital products, protecting rightful ownership has become difficult. Currently, there are two schemes to protect data on the Internet. The first scheme is the traditional cryptography where the important data or secret is to be encrypted by a special process before being transmitted on the Internet. This scheme requires much computational process and time to encrypt or decrypt. On the other hand, the second scheme is steganography where the important message or secret is hidden in the digital media. The hidden data is not perceptible by the human visual system (HVS). The digital watermarking technique is an application of steganography (Chang, Huang, & Chen, 2000; Chen, Chang, & Huang 2001). In order to safeguard copyrights and rightful ownerships, a representative logo or watermark could be hidden in the image or media that is to be protected. The hidden data can be recovered and used as proof of rightful ownership. The watermarking schemes can be grouped into three kinds, largely, dependent on its application. They use the fragile watermark, semi-fragile watermark, and robust watermark, respectively (Fabien, Ross, & Markus, 1999). Fragile watermarks are easily corrupted when the watermarked image is compressed or tampered with. Semi-fragile watermarks can sustain attacks from normal image processing, but are not robust against malicious tampering. Fragile and semi-fragile watermarks are restricted in its use for image authentication and integrity attestation (Fridrich,2002; Fridrich, Memon, & Goljan, 2000). For the robust watermarking, it is always applied in ownership verification and copyright protection (Fridrich, Baldoza, & Simard, 1998; Huang, Wang, & Pan, 2002; Lu, Xu, & Sun, 2005; Solanki, Jacobsoen, Madhow, Manjunath, & Chandrasekaran, 2004). Some basic conditions must be followed: (1) Invisibility: the watermarked image must look similar to its original and any difference invisible to the human visual system. (2) Undetectable: the watermark embedded in the image must not be easily detectable by computing processes or statistical methods. (3) Safety: watermark is encrypted and if accessed by a hacker; cannot be removed or tampered with. (4) Robustness: the watermark is able to withstand normal and/or illegal manipulations, such as compression, blurring, sharpening, cropping, rotations and more. The retrieved watermark is perceptible even after these processes. (5) Independence: the watermark can be retrieved without the original image. Last but not the least, (6) Efficiency: the watermarked image should not require large storage and must also allow for a comparable-sized watermark to be hidden in the media. The proposed method is a VQ-based watermark technique that depends on the structure of a tree growth for grouping the codebook. The scheme is robust. That is, the watermark is irremovable and also can withstand normal compression process, tampering by compression or other malicious attacks. After these attacks, the watermark must be recovered with comparable perceptibility and useful in providing proof of rightful ownerships.
A Novel Fragile Watermark Algorithm for Detecting Speech Tampering
