ABSTRACT

This chapter discusses different types of watermarking attacks and available tools. It addresses watermarking attack schemes into four classes: image processing attacks, geometric transformation, cryptographic attacks and protocol attacks. The idea behind blurring is to decrease the magnitude of high-frequency components. The low-pass filter in the frequency domain is equivalent to the mountain shape in the spatial domain. Image scaling can be of one of two types: uniform or non-uniform. Uniform scaling uses the same scaling factors in horizontal and vertical directions. However, non-uniform scaling applies different scaling factors in horizontal and vertical directions. Image warping performs a pixel-by-pixel remapping of an input image to an output image. It requires an integer-sampled "remapping image" that is the same size as the output image. Templates are used as a synchronization pattern to estimate the applied affine transformation. Many modern schemes use a template in the frequency domain to extract the watermark.