Image sets used in the development of a connected auto-encoders based approach to separate mixed X-radiographs from double-sided paintings
- 1. National Gallery, London
Description
The following sets of images were used during the development of an algorithm (described in the publication detailed below) designed to separate the mixed X-radiographs from double-sided paintings into two hypothetical X-ray images corresponding to each side of the painting, when visible images of the two sides of the painting are available.
The images sets are taken from a painting that is only painted on one side and were used to assess the regularization parameters associated with the separation approach. The details are taken from the visible image and the X-radiograph of Anthony van Dyck’s painting Lady Elizabeth Thimbelby and Dorothy, Viscountess Andover dated to about 1635 and now in the collection of the National Gallery in London (NG6437). See https://www.nationalgallery.org.uk/paintings/anthony-van-dyck-lady-elizabeth-thimbelby-and-her-sister for further details of the painting.
The code can be downloaded from: https://github.com/ART-ICT/Xray_Separation_2RGB and the algorithm is described in W. Pu, B. Sober, N. Daly, C. Zhou, Z. Sabetsarvestani, C. Higgitt, I. Daubechies and M. Rodrigues, ‘Image Separation with Side Information: A Connected Auto-Encoders Based Approach’, Transactions on Image Processing, 2023
All images © The National Gallery, London
Datasets available:
NG6437_vis_800pixel_230502.tif: 800 pixel thumbnail visible image of the entire painting showing the location of the two details used for the algorithm development. This image is derived from a visible image of the whole painting acquired 25 November 2019 (Original file: N-6437-00-000041.tif; 6272 x 5940 pixels).
NG6437_xray_800pixel_230502.tif: 800 pixel thumbnail image of the X-radiograph of the entire painting showing the location of the two details used for the algorithm development. This image is derived from the composite X-radiography of the whole painting created by mosaicking digital scans of the individual sheets of film and then registering the resulting image to the high resolution visible image described above (Original file: N-6437-00-000049.tif; 36847 x 32516 pixels).
NG6437_vis_crop_01_230502.tif: 1543 x 2078 pixel detail taken from the high resolution visible image described above.
NG6437_xray_crop_01_230502.tif: 1543 x 2078 pixel detail of the X-radiograph corresponding to NG6437_vis_crop_01_230502.tif. The X-ray images were acquired using sheets of film (27 November 2019) and 16-bit digital scans were then produced (original files: N-6437-00-000047-009 and -014 (each 9539 x 7199 pixels), processed 28 January 2020). This crop is an 8-bit composite image of 2 X-ray plates that had been manually registered to the high resolution visible image described above using Adobe Photoshop.
NG6437_vis_crop_02_230502.tif: 1562 x 2023 pixel detail taken from the high resolution visible image described above.
NG6437_xray_crop_02_230502.tif: 1543 x 2078 pixel detail of the X-radiograph corresponding to NG6437_vis_crop_02_230502.tif. The X-ray images were acquired using sheets of film (27 November 2019) and 16-bit digital scans were then produced (original files: N-6437-00-000047-002 and -007 (each 9539 x 7199 pixels), processed 28 January 2020). This crop is an 8-bit composite image of 2 X-ray plates that had been manually registered to the high resolution visible image described above using Adobe Photoshop.
NG6437_vis_crop_03_230502.tif: 2088 x 2088 pixel detail taken from the high resolution visible image described above.
NG6437_xray_crop_03_230502.tif: 2088 x 2088 pixel detail taken from the composite X-radiograph described above corresponding to NG6437_vis_crop_03_230502.tif.
NG6437_vis_crop_04_230502.tif: 2088 x 2088 pixel detail taken from the high resolution visible image described above.
NG6437_xray_crop_04_230502.tif: 2088 x 2088 pixel detail taken from the composite X-radiograph described above corresponding to NG6437_vis_crop_04_230502.tif.
Notes
Files
NG6437_vis_800pixel_230502.tif
Files
(98.9 MB)
| Name | Size | Download all |
|---|---|---|
|
md5:64c4043dd7ab7d88ca543b310f594264
|
1.7 MB | Preview Download |
|
md5:699e655a4786ed73233e2b607ef6336f
|
9.6 MB | Preview Download |
|
md5:769435b5f313cd2527ab7156adf0a60f
|
9.5 MB | Preview Download |
|
md5:f395f58f4645531f03d99cec57fd63a4
|
13.1 MB | Preview Download |
|
md5:c30fd8e9ee4f19dca78044f970a12a2f
|
13.1 MB | Preview Download |
|
md5:da26e88b78f9689e6b7d997ddfd71691
|
2.2 MB | Preview Download |
|
md5:5938c3096654f0e52528f09b70373553
|
9.6 MB | Preview Download |
|
md5:bc0fd1267e2c8d43f32f5347a0864113
|
9.5 MB | Preview Download |
|
md5:acccde9653e4c4ec138b1fc193512612
|
17.4 MB | Preview Download |
|
md5:9632f9cecae697b8b018679699225ac6
|
13.1 MB | Preview Download |
Additional details
Related works
- Is supplemented by
- Software: https://github.com/ART-ICT/Xray_Separation_2RGB (URL)
Funding
- UK Research and Innovation
- ARTICT | Art Through the ICT Lens: Big Data Processing Tools to Support the Technical Study, Preservation and Conservation of Old Master Paintings EP/R032785/1