An Improved U-Net for Watermark Removal

Convolutional neural networks (CNNs) with different layers have performed excellent results in watermark removal. However, how to extract robust and effective features via CNNs of black box removal is very important. In this paper, we propose an improved U-net (IWRU-net). Taking the robustness obtained information into account, a serial architecture designed facilitate useful for guaranteeing performance problem long-term dependency U-nets based simple components are integrated more salient hierarchical addressing problems. To increase adaptability IWRU-net real world, use randomly distributed blind watermarks implement model. The experiment illustrate that proposed method superior other popular methods terms quantitative qualitative evaluations.