Attack Type Agnostic Perceptual Enhancement of Adversarial Images
This work addresses a practical issue for users of CAPTCHAs and similar systems by improving human readability of adversarial images, though it is incremental as it builds on existing attacks.
The paper tackles the problem of degraded perceptual quality in adversarial images, which makes them difficult for humans to classify, by proposing an attack-agnostic enhancement method that reduces Euclidean distances by an average of 22% while maintaining adversarial performance.
Adversarial images are samples that are intentionally modified to deceive machine learning systems. They are widely used in applications such as CAPTHAs to help distinguish legitimate human users from bots. However, the noise introduced during the adversarial image generation process degrades the perceptual quality and introduces artificial colours; making it also difficult for humans to classify images and recognise objects. In this letter, we propose a method to enhance the perceptual quality of these adversarial images. The proposed method is attack type agnostic and could be used in association with the existing attacks in the literature. Our experiments show that the generated adversarial images have lower Euclidean distance values while maintaining the same adversarial attack performance. Distances are reduced by 5.88% to 41.27% with an average reduction of 22% over the different attack and network types.