DCT-based Watson's imperceptive watermarking model resisted downsampling attack is presented.The downsampling image is obtained by quasi-convoluted downsampling and the region

the length and the strength of watermark sequence is adaptively embedded in the downsampling image by exploiting the luminance characteristic and a perceptual distortion constraint

which greatly improves the perceptual quality and robustness of up-sampling watermarked image.Simulation results show that the proposed method can achieve a better robustness to the white noise and video downsampling and the original image is not required for watermark detection.