A New Pre-processing Approach to Improve DCT-Based Watermarkings Extraction

Discrimination between watermark and host image is one of primary problems in DCT-based watermarking algorithms, especially when attack occurred on watermarked image. In this paper, a new pre-processing approach is presented to improve watermark extracting performance of DCT-based watermarking schemes. This method is based on adding noise magnifying and enhancement operations as beginning steps to watermark extraction procedure. First, the watermark information is treated as noise for the watermarked image in its spatial domain. Then, the noise is magnified before detection. Next, enhancement operations are done by applying combination of two filter; sharpening and Laplacian of Gaussian, respectively. To show effectiveness of proposed method, a typical DCT method is used and proposed pre-processing package is added to this method. Experimental results show that extraction performance of DCT-based watermarking algorithm is improved by applying our pre-processing scheme before watermark extraction procedure.