Reversible attack based on local visible adversarial perturbation

摘要

Adding perturbation to images can mislead classification models to produce incorrect results. Based on this, research has exploited adversarial perturbation to protect private images from retrieval by malicious intelligent models. However, adding adversarial perturbation to images destroys the original data, making images useless in digital forensics and other fields. To prevent illegal or unauthorized access to sensitive image data such as human faces without impeding legitimate users, the use of reversible adversarial attack techniques is becoming more widely investigated, where the original image can be recovered from its reversible adversarial examples. However, existing reversible adversarial attack methods are designed for traditional imperceptible adversarial perturbation and ignore the local visible adversarial perturbation. In this paper, we propose a new method for generating reversible adversarial examples based on local visible adversarial perturbation. The information needed for image recovery is embedded into the area beyond the adversarial patch by the reversible data hiding technique. To reduce image distortion, lossless compression and the B-R-G (blue-red-green) embedding principle are adopted. Experiments on CIFAR-10 and ImageNet datasets show that the proposed method can restore the original images error-free while ensuring good attack performance.

关键词