Image high-capacity robust watermarking method based on wavelet neural network

Aiming at the limitation of the prior art, the invention provides the image high-capacity robust watermarking method based on the wavelet neural network, the network structure is improved, and the ordinary convolution up-sampling and down-sampling operations are respectively replaced by the wavelet transform operation and the wavelet inverse transform operation. The improvement enables the network to retain more high-frequency information during down-sampling and to be more robust to noise. 本发明针对现有技术的局限性，提出了一种基于小波神经网络的图像大容量鲁棒水印方法，在网络结构上进行了改进，采用了小波变换操作和小波逆变换操作来分别替代普通的卷积上采样和下采样操作，这种改进能使得网络在下采样时能够保留更多的高频信息，以及对噪声更加鲁棒。