Table 1 Attack types with varying parameters.
From: Robust zero-watermarking for color images using hybrid deep learning models and encryption
Types of attacks | Parametric description of attacks |
|---|---|
Filtering | Average filtering using window sizes of 3 × 3, 5 × 5, and 7 × 7 |
Gaussian filtering using window sizes of 3 × 3, 5 × 5, and 9 × 9 | |
Median filtering using window sizes of 3 × 3, 5 × 5, and 7 × 7 | |
Wiener filtering using window sizes of 3 × 3, 5 × 5, and 7 × 7 | |
Geometric transform | Rotation (1°, 3°, 5°, 10°, 50°) using nearest, bilinear, and bicubic interpolation |
Scaling (0.25, 0.5, 2.0, 4.0) using nearest, bilinear, and bicubic interpolation | |
JPEG compression | quality factor = 5, 10, 20, 30, 40, 50, 60, 70, 80, 90 |
Brightness adjustment | Histogram equalization |
Noise | Gaussian noise with parameters 0.1, 0.2, 0.3, and 0.5 |
Salt and pepper noise with parameters 0.1, 0.2, 0.3, and 0.5 | |
Translation attack | factor = 10, 20, 40, 60 |
Conventional combined attacks | Median filter (5 × 5) + Gaussian noise (0.3) |
Median filter (5 × 5) + Salt & pepper noise (0.3) | |
Wiener filter (5 × 5) + Salt & pepper noise (0.3) | |
Median filter (5 × 5) + JPEG Compression (90) | |
Gaussian noise (0.3) + JPEG Compression (90) | |
Salt & pepper noise (0.3) + JPEG Compression (90) | |
Rotation (2°) + JPEG Compression (10) | |
Scaling (0.5) + JPEG Compression (90) | |
Center cropping attack | 32 × 32, 64 × 64, 128 × 128 |
Top left cropping attack | 32 × 32, 64 × 64, 128 × 128 |
