Embedding Large Watermarks In Images Using Discrete Fractional Cosine Transform

The amount of information that can be hidden in an image without any visually noticeable difference and retrieval of this embedded information are the difficulties in watermarking techniques. In this paper we present a novel algorithm for embedding and extracting large digital watermarks in images. This method is inspired by Cox’s watermarking strategy. Unlike embedding an image or binary watermarks that are smaller than the host image being watermarked, in our algorithm we demonstrate by embedding an audio watermark which is much larger than the host image to be watermarked. The size of the watermarked image increases only by a fraction after watermarking. The positions where the watermarks are to be embedded, intermediate frequency coefficients and the sequence of the transformation orders comprise the key. The experimental results have shown that the proposed algorithm is of good imperceptibility and security. It is also robust against common attacks like rotation, resizing, filtering and even cropping.