In this paper, an efficient blind watermark detection scheme using DWT coefficients is proposed. The embedding scheme is multiplicative and done at second level of DWT decomposition by optimum choice of the embedding strength. The detection is based on the maximum likelihood ratio method. Neyman-Pearson criterion is used to minimize the missed detection probability subject to a fixed false alarm probability. The DWT coefficients are assumed to be modeled using the Laplacian distribution. In , the watermark is embedded in the vertical, horizontal and diagonal subbands in the III level. In this paper, we propose to embed the watermark only in the diagonal subband in level II, which results in better imperceptibility, robustness and capacity.