These are the results accompanying the paper "Virtual Restoration of the Ghent Altarpiece Using Crack Detection and Inpainting" submitted to the conference Advanced Concepts for Intelligent Vision Systems (ACIVS) 2011.
The crack maps used as input for inpainting are obtained as explained in Section 2 of the paper. Inpainting methods used, apart from the proposed one, are:
[1] Perona, P., Malik, J.: Scale-space and edge detection using anisotropic diffusion. IEEE Trans. on Pattern Anal. and Machine Intel., vol. 12, no. 7, 629-639 (1990)
[2] Criminisi, A., Perez, P., Toyama, K.: Region filling and object removal by exemplar-based image inpainting. IEEE Trans. Image Proc., vol. 13, no. 9, 1200-1212 (2004)
[3] Komodakis, N., Tziritas, G.: Image completion using efficient belief propagation via priority scheduling and dynamic pruning. IEEE Trans. Image Proc., vol. 16, no. 11, 2649-2661 (2007)
All the images are downscaled by a factor of 2, except from the last example, which is downscaled by the factor of 4. Click on the thumbnail for an enlarged image.
Comparison of different inpainting methods on the small part of the image of a jewel:
|
|
|
|
|
Influence of white borders around the cracks on the inpainting result. First row: results are obtained with the crack map without white borders and bright cracks. Second row: results are obtained with the crack map with white borders and bright cracks.
|
|
|
|||
|
|
|
|
|
|
|||
|
|
|
Results of the proposed constrained candidate selection from Section 3.2 as an improvement for patch-based inpainting:
|
|
|
|
Results of the proposed constrained candidate selection with adaptive patch size on the whole book:
|
|
|