A Robust Digital Watermarking using Discrete Wavelet Transform in Chrominance Channel Komwit Surachat Information and Communication Technology Programme, Faculty of Science, Prince of Songkla University, Thailand Email: komwit.s@psu.ac.th Abstract—A wavelet-based digital watermarking scheme is presented in this paper. The watermark signal is embedded into HLn of LLn-1 transformed sub-band of original chrominance channel host image using coefficients scaling technique. In the extraction process, the average filter is applied to predict original dwt coefficients in HLn-1 sub- band. Then, the watermark logo is recovered by subtracting the coefficients predicted from the embedded coefficients, without original image needed. The obtained results show significant improvement in terms of quality and fidelity of extracted signal. Also, the improvements in terms of robustness by the proposed method work effectively against various common image-processing-based attacks especially against compression attacks. Index Terms—discrete wavelet transform, digital watermarking, chrominance channel, image processing based attack I. INTRODUCTION The copyright of digital media protection becomes an important topic nowadays, due to one main characteristic of digital multimedia which is the ease of copying and redistributing without losing quality. To resolve the copyright protection problem for multimedia data, many watermarking schemes are proposed and discussed. We can classify watermarking methods into frequency and spatial domain based watermarking. In the spatial domain, watermark embedding can be accomplished easily by modifying the image pixels directly. For instance, [1], [2] and [3] also proposed the watermarking schemes in spatial domain by using spatial correlation of colors, RST-resistant method and independent component analysis, respectively. However, there are many arguments about robustness against compression attacks e.g. JPEG2000 and JPEG compression standard. In contrast, in the frequency domain based approach, it is obvious that the robustness against compression attack is much better than spatial domain based. Moreover, frequency domain watermarking strongly helps increasing the imperceptibility, security, and robustness. Therefore, presently, most of image watermarking methods are in the transform domain. For Manuscript received October10, 2012; revised December 22, 2012. This research work is supported by the Information and Communication Technology Programme, Faculty of Science, Prince of Songkla University. example, M. K. Samee et al. [4] presented reversible watermarking scheme for images by using CMDA based in wavelet domain. In addition, [5] and [6] proposed watermarking algorithms in frequency domain using discrete wavelet transform. They applied discrete fractional Fourier transformation DWT and Region of Interest (ROI) technique, respectively. In this paper, we present a wavelet-based digital watermarking scheme by applying some spatial domain techniques to improve the performance of extracted watermark signal. The coefficient scaling method is proposed to adjust area before embedding which will be explained more in section 2. Then, in section 3, the experimental results are shown and discussed. The conclusion is finally drawn in section 5. II. PROPOSED WATERMARKING SCHEME This proposed watermarking scheme consists of two processes those are the embedding process and the extraction process. In addition, we also present some techniques to transform the color channel and the sizes of watermark image and host image before performing the operations. The detail of the embedding process and sub- processes are given next. A. Pre-processing Operation of Host Image The host image is first pre-processed before embedding with the following steps; Step 1: The original RGB host image I with n by n pixels is converted to YC b C r color space, the chrominance channel, C r is selected to use in this process.   240 16 , 1 , 1 | ) , ( ) , (        j i r Cr n j n i Cr C j i Then, the host image C r is decomposed into n levels using discrete wavelet transform. The HL n of LL n-1 transformed sub-band is selected for watermarking embedding.     1 , , ,   n n n n n LL dwt r b q a (1) where n n n n r b q a , , , is wavelet coefficient value of LL n ,HL n LH n and HH n sub-band, respectively. Step 2: The coefficient values q n(i,j) is scaled between 0-255 by using following equations;   ) , ( ) , ( min ' j i n n j i n q q q    (2) 99 ©2013 Engineering and Technology Publishing doi: 10.12720/joace.1.2.99-102 Journal of Automation and Control Engineering, Vol. 1, No. 2, June 2013