Benjamin_Loison commented on issue Benjamin_Loison/Robust_image_source_identifica…#25 2024-03-29 13:39:01 +01:00
Estimating fake PRNU on noise-free images

I start having the feeling that the RMS inconsistency is because of the PRNU estimation background which can not converge to 0 as it is just adding residual scene.

81852d5dc6 Make Gaussian noise in getPrnuShownAsSuch only on text
np.random.normal scale specify a given proportion of noise being in a given interval?
ec7c5c6688 Add gaussianNoise to getPrnuShownAsSuch but applies on the whole PRNU image
Benjamin_Loison commented on issue Benjamin_Loison/Robust_image_source_identifica…#27 2024-03-29 13:12:07 +01:00
Generate arbitrary shape PRNU shown as such

2d96bdc225092efb6cc09479cb8696c931f7a522

2d96bdc225 Not correct generalized SPLIT_N_X_N due to imageWithoutPrnuNpArrayTile
96bbd50a3b Generalize with a partial image SPLIT_N_X_N
Benjamin_Loison commented on issue Benjamin_Loison/Robust_image_source_identifica…#27 2024-03-29 12:45:41 +01:00
Generate arbitrary shape PRNU shown as such

Should also be able to Gaussian noise the generated text.

4bf9ef8206 Add splitNxN compatible with 1 and 4
Benjamin_Loison commented on issue Benjamin_Loison/Robust_image_source_identifica…#27 2024-03-29 12:43:13 +01:00
Generate arbitrary shape PRNU shown as such
import os
from PIL import Image, ImageFont, ImageDraw 

# Supports `WIDTH` > `HEIGHT` and conversely.
WIDTH = 1_000
HEIGHT = 1_000
TEXT = 'PRNU'

img = Image.new('L', (WIDTH,
60eceb575e Simplify grayscale rendering
b375acbb3a Use v{min,max} for enforcing matplotlib colormap
Compare 2 commits »
Benjamin_Loison commented on issue Benjamin_Loison/Robust_image_source_identifica…#25 2024-03-29 11:26:05 +01:00
Estimating fake PRNU on noise-free images

How to ease visualizing the values of this rendering of image of floats?

5fa61f7ff8 Make PRNU compatible with 4x4 split
Benjamin_Loison commented on issue Benjamin_Loison/Robust_image_source_identifica…#25 2024-03-29 11:22:50 +01:00
Estimating fake PRNU on noise-free images

Concerning the RMS, maybe it is due to residual scene and with more images we would have a many pixels, hence important, less significant as nearer to 0.

Benjamin_Loison commented on issue Benjamin_Loison/Robust_image_source_identifica…#25 2024-03-29 02:03:38 +01:00
Estimating fake PRNU on noise-free images

If want to move on Gaussian noise as PRNU, then have to understand RMS unexpected results.

8351e46437 Comparison of an image without and with Gaussian noise and PRNU
Benjamin_Loison commented on issue Benjamin_Loison/Robust_image_source_identifica…#25 2024-03-29 01:48:12 +01:00
Estimating fake PRNU on noise-free images

Let us not consider this RMS inconsistency but admit that the visual clear PRNU is fine, so let us add Gaussian noise.