This is the mostbasic but non-trivial requirement that all watermarking schemes should meet. For most applications, the watermark should also be robust to some intentional/ unintentional attacks. To provide a robust watermark, a good strategy is to embed the watermark signal into the significant portion of the host signal.

Visible watermarks, as the name says, are visual patterns, like logos, which are inserted into or overlaid on images (or video), very similar to visible paper watermarks. Visible watermarks are mainly applied to images, for example, to visibly mark preview images available in image databases or on the World Wide
Web in order to prevent people from commercial use of such images.

Fragile watermarks do not survive lossy transformations since their purpose is tamper detection in the original signal. There are many effective ways to insert a fragile watermark into digital content while preserving the imperceptibility requirement [8]. Placing the watermark information into the perceptually insignificant portions of the data guarantees imperceptibility and provides fragile marking
capabilities.

For security applications and copyright protection, robust watermarking techniques have been proposed [9-12].

Here the technical challenge is to provide transparency and robustness that are conflicting requirements. Ideally, an effective, robust watermarking scheme provides a mark that can only be removed when the original content is destroyed as well.

Spatial and Spectral Watermark

The watermark in a signal can be embedded in the spatial domain [13] or in transform domain like the discrete cosine transform (DCT) domain, discrete Fourier transform (DFT) or wavelet domain [14-16]. For image watermarking the signal embedding is done by signal-adaptive (i.e. scaled) addition, mostly to the luminance channel alone, but sometimes color channels are also included. It is often claimed that embedding in the transform (mostly DCT or wavelet) domain is advantageous in terms of visibility and security. However, while some authors argue that the watermarks should be embedded into low frequencies, others argue that they should rather be embedded into the medium or high frequencies. In fact, it has been shown that for maximum robustness, signal adaptive watermarks should be embedded into the same spectral components
that the host data already populate. For images and video, these are typically the low frequencies.