Estimate PRNU on easy real images #29

New Issue

Benjamin_Loison · 2024-03-30T00:07:51+01:00

Benjamin_Loison commented

2024-03-30 00:07:51 +01:00

#31: Can for instance for a given device split into two groups and compare both estimated PRNUs and how they evolved when consider more and more images to estimate them within each group.
#30: Otherwise can estimate PRNUs for each device of a dataset and then use new images of these devices to estimate what device they are associated to and can monitor the performances thanks to the accuracy.

I would expect a software showing me not demosaiced RGB but it is not the case of darktable, GIMP and vpv by default for .nef and .tif...

Related to #3 and #49.

- #31: Can for instance for a given device split into two groups and compare both estimated PRNUs and how they evolved when consider more and more images to estimate them within each group. - #30: Otherwise can estimate PRNUs for each device of a dataset and then use new images of these devices to estimate what device they are associated to and can monitor the performances thanks to the accuracy. I would expect a software showing me not demosaiced RGB but it is not the case of darktable, GIMP and vpv by default for `.nef` and `.tif`... Related to #3 and #49.

Benjamin_Loison added the

enhancement

high priority

epic

labels 2024-03-30 00:07:51 +01:00

Benjamin_Loison referenced this issue

2024-03-30 00:08:27 +01:00

Estimating fake PRNU on noise-free images #25

Benjamin_Loison pinned this 2024-03-30 00:08:33 +01:00

Benjamin_Loison referenced this issue from a commit

2024-04-15 16:06:10 +02:00

#29: Add `Nikon D7000` 4946x3278 PRNU estimation from NEF to PNG images

Benjamin_Loison commented

2024-04-15 16:14:53 +02:00

Tried estimating PRNU on not flat-field PNG from NEF images but I get the following for Nikon D7000 4946x3278 4,651 images:

100%|█████████████████████████████████████| 8156/8156 [00:02<00:00, 3852.64it/s]
100%|█████████████████████████████████████| 4651/4651 [5:47:23<00:00,  4.48s/it]

not being really interesting it seems as almost all pixels are black.

Related to #48.

Tried estimating PRNU on not flat-field PNG from NEF images but I get the following for `Nikon D7000` 4946x3278 4,651 images: ``` 100%|█████████████████████████████████████| 8156/8156 [00:02<00:00, 3852.64it/s] 100%|█████████████████████████████████████| 4651/4651 [5:47:23<00:00, 4.48s/it] ``` ![np_array](/attachments/b1fd7565-218b-4bed-9331-d585d0435862) not being really interesting it seems as almost all pixels are black. Related to #48.

np_array.png

70 KiB

Benjamin_Loison unpinned this 2024-05-03 13:47:20 +02:00

Sign in to join this conversation.