Two‐stage visible watermark removal architecture based on deep learning

Abstract

With the rapid development of the Internet, watermarks are widely used in images to protect copyright. This implies that the robustness of watermark is very important. In recent years, there have been some studies to evaluate watermark performance by removing the watermark. Among them, some methods need to mark the watermark position in advance, and some require multiple images with the same watermark. Moreover, when the colour of thewatermark is similar to that of the background, the existing methods can hardly remove the watermark from the watermarked image. In the proposed work, the authors presented a watermark removal structure consisting of watermark extraction and image inpainting to address the aforementioned issues. In particular, the extraction network is used to extract the watermark in the watermarked image, and the inpainting network is used to inpainting image for a better watermark removal image, respectively. Finally, the authors train and test the developed network architecture by constructing two data sets, i.e. white watermarked image data set (WW-data set) and colour watermarked image data set (CW-data set). The proposed method not only has better performance on the WW-data set than the current latest methods (on the CW-data set, other methods have almost failed) but also effectively removes the watermarks.

A Review of Document Image Enhancement Based on Document Degradation Problem

“…However, it is necessary to set the parameters manually, resulting in the need to manually set the results of the algorithm according to the features of different images, which leads to a decrease in model efficiency when facing multiple datasets. In 2020, Jiang et al proposed a visible watermark removal network architecture [50] based on conditional generative adversarial networks (CGAN) [51] and least squares generative adversarial networks (LSGAN) [52]. As shown in Figure 11, the proposed model can automatically detect the watermark location without manual checking, which improves the model efficiency, but the generalization ability of the model is weak.…”

Survey on deep learning applications in digital image security

“…The generated watermark-less image had photorealistic quality but not good performance in standard quantitative evaluation metrics such as PSNR. Jiang et al 181 presented a watermark removal structure consisting of a watermark extraction network that removed the watermark in the watermarked image and an image inpainting network that inpainted the image for a watermark-less image. The two networks were trained on the PASCAL VOC2012 and places2 182 datasets.…”

A novel ensemble framework based on CNN models and Swin transformer for cervical cytology image classification

“…A Transformer network is a cutting-edge architecture which has been effectively used in the medical image process. A vision transformer (VIT) transfers the transformer model to computer vision tasks, offering a self-attention mechanismbased image investigation framework [7]. While convolutional neural networks (CNNs) have long dominated modeling in computer vision and are widely adopted for extracting features in particular regional areas, they are not capable to capture the contextual connections among image features on a global scale and frequently overlook critical details.…”