Robust PDF Watermarking against Print–Scan Attack

View full-text article in PMC

. 2023 Aug 23;23(17):7365. doi: 10.3390/s23177365

Algorithm 2 Extraction algorithm

Input: Original watermarked PDF file and watermarked PDF file

Output: Extracted watermark sequence

1:
Convert the original PDF file and the watermarked PDF file into a series of images, and perform subsequent operations on each original image $I$ and the watermarked image $I_{W}$ . Map the image $I$ and $I_{W}$ to be detected from RGB space to YCbCr space, decompose the $Y$ components, mark it as $I_{Y}$ and $I_{Y}^{'}$ , and convert them into binary text images $I_{b}$ and $I_{b}^{'}$ .
2:
for each original image $I_{b}$ and corresponding watermarked image $I_{b}^{'}$ do
3:
Obtain the set of sub-blocks ${B_{i} | i = 1, 2, \dots, K, K = 2 P}$ , and ${B_{i}^{'} | i = 1, 2, \dots, K, K = 2 P}$ , where P is the number of rows.
4:
for each $B_{i}$ do
5:
Adjust its corresponding $B_{i}^{'}$ to ensure that the pixels of $B_{i}$ and $B_{i}^{'}$ correspond one by one.
6:
Calculate the number of black pixels in each row of $B_{i}$ , find the row with the highest number of black pixels, and mark it as $r_{l o n g}$ .
7:
for each column in L do
8:
Find the white pixel closest to $r_{l o n g}$ , and mark it as $I_{b} (x, y)$ .
9:
Extract pixel $I_{b}^{'} (x, y)$ from the watermarked image $I_{b}^{'}$ .
10:
Calculate the difference $m_{1}^{(i)}$ .
11:
end for
12:
Obtain sets $M_{1} = {m_{1}^{(i)} | i = 1, 2, \dots, N}$ .
13:
Extract watermark bit according to the number of positive and negative numbers in $M_{1}$ .
14:
end for
15:
Extract all watermark bits and obtain the watermark string (length N) extracted from the current image.
16:
end for
17:
Overlay and average all extracted watermark strings to obtain the final extracted watermarks $w_{1}^{'}, w_{2}^{'}, \dots, w_{N}^{'}$ .