Scalability Based Blind Watermarking in Frequency Domain Conscious Robustness with Lossless Watermark Extraction

In this paper, a scalability based blind image watermarking using third-level discrete wavelet transform (DWT) is proposed. The wavelet coefficients of the host are replaced by a watermark bit that contains the watermark information. Initially, for each coefficient, a binary sequence is estimated by using a quantized binary tree. The states of the binary sequences are estimated which acts as the `n' bit present state. Using the `n' bit present state and watermark bit, the `n' bit next state is calculated. The first `n' bit of wavelet coefficient is replaced by the `n' bit next state which acts as watermark coefficient. The watermarked image can be estimated after obtaining the inverse discrete wavelet transform (IDWT) for the watermark coefficients. During the watermark extraction process, the watermarked coefficients are estimated using the wavelet transform without using quantization based binary tree formation. The watermark information can be decoded from the state of first `n' bit of the coefficient. The experimental results of the proposed system were tested on different test images with the metrics such as embedding distortion rate, data capacity, peak signal to noise ratio, structural similarity index measurement and hamming distance. The simulation results verify that the proposed system outperforms from the conventional blind image watermarking techniques.