Two-branch Multi-scale Deep Neural Network for Generalized Document Recapture Attack Detection

TL;DR

This paper introduces a two-branch deep neural network that enhances the detection of recaptured document images by leveraging frequency filtering and multi-scale attention, improving generalization across scenarios.

Contribution

The paper proposes a novel two-branch deep neural network with frequency filtering and multi-scale cross-attention to better detect recaptured document images and address overfitting issues.

Findings

Outperforms state-of-the-art methods in generalization across scenarios

Effective in mining recapture artifacts with frequency domain analysis

Improves detection accuracy for personal document recapture attacks

Abstract

The image recapture attack is an effective image manipulation method to erase certain forensic traces, and when targeting on personal document images, it poses a great threat to the security of e-commerce and other web applications. Considering the current learning-based methods suffer from serious overfitting problem, in this paper, we propose a novel two-branch deep neural network by mining better generalized recapture artifacts with a designed frequency filter bank and multi-scale cross-attention fusion module. In the extensive experiment, we show that our method can achieve better generalization capability compared with state-of-the-art techniques on different scenarios.