«

PDF Dateien pixelbasiert miteinander vergleichen

2025-11-23, 19:52

Ausgehend von einer Anfrage auf der Dante Beraterkreis Mailingliste hier einige Ideen, wie man pixelbasiert PDF-Dateien miteinander vergleichen kann.

Schritt 1: Umwandlung ins PNG Format. Ich nutze dazu pdftoppm aus dem Poppler-Paket.


pdftoppm -png 1.pdf > 1.png

das gleiche dann nochmal für die zweite Datei.

Dann mit ein wenig Python-Magie die beiden PNGs vergleichen:

 
from PIL import Image, ImageChops # generic
import numpy as np # numerical
from skimage.metrics import structural_similarity as ssim
import cv2
 
# Load the images
img1 = Image.open("1.png")
img2 = Image.open("2.png")
 
# Compute absolute difference
diff = ImageChops.difference(img1, img2)
 
# Save or show the difference image
diff.save("difference.png")
diff.show()
 
 
# Load PNGs into numpy arrays
img1 = np.array(Image.open("1.png"))
img2 = np.array(Image.open("2.png"))
 
# Compute pixel-wise absolute difference
diff = np.abs(img1.astype(int) - img2.astype(int))
 
# Save result
Image.fromarray(diff.astype(np.uint8)).save("difference.png")
# Or compute some stats:
print("Mean difference:", diff.mean())
print("Max difference:", diff.max())
 
 
# Convert to grayscale
gray1 = cv2.cvtColor(img1, cv2.COLOR_BGR2GRAY)
gray2 = cv2.cvtColor(img2, cv2.COLOR_BGR2GRAY)
 
score, diff = ssim(gray1, gray2, full=True)
print("SSIM:", score)
 
# diff is in [0,1]; scale to 0-255 to view as an image
diff = (diff * 255).astype("uint8")
cv2.imwrite("ssim_diff.png", diff)
Category: LaTeX, Tipps & Tricks