Adversarial neural network-based watermark removal method, device and apparatus and readable medium

Abstract

The invention provides an adversarial neural network-based watermark removal method, device and apparatus and a readable medium. The method comprises a training step of training an adversarial neuralnetwork on the basis of sample images so as to obtain an adversarial neural network, and a removal step of inputting images with watermarks into the adversarial neural network so as to obtain images without watermarks. According to the method, device and apparatus and the readable medium, the adversarial neural network with residual blocks is designed; the adversarial neural network comprises a generator network and a discriminator network; through the generator network in the adversarial neural network specific watermarks can be rapidly and efficiently removed in batches; and through the discriminator network, judgement is carried out determine whether the watermark removal effect achieves requirements or not, and if the judging result is positive, images are output.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method, device, device and readable medium for removing watermarks based on an adversarial neural network. Background technique [0002] Watermarks are often added to many images in a consistent manner. When it is very challenging to automatically remove watermarks from a single image, visual watermarks usually contain complex structures such as thin lines and shadows, making them more difficult to remove. For dealing with the problem of watermarking, there are several current techniques. [0003] 1. Manual processing by image processing software. For example, use photoshop to remove the watermark. [0004] 2. Google demonstrated a computer algorithm that can automatically obtain image copyright protection and remove watermarks, so that users can access clear images without barriers. A generalized multi-image concealment algorithm is proposed that takes as input an ...

AI Technical Summary

Problems solved by technology

These techniques operate on a single image, require the user to manually mark watermarked regions, and cannot handle large watermarked regions

[0006] It can be seen that the shortcomings of the existing technology are: it is impossible to perform large-scale automatic watermark removal operations, and the number of watermarks that appear on a single picture cannot be too large, and the watermark area needs to be marked manually

Images