TY - CHAP
T1 - Crack Detection in Oil Paintings Using Morphological Filters and K-SVD Algorithm
AU - Rucoba-Calderón, Carla
AU - Ramos, Efrain
AU - Gutiérrez-Cárdenas, Juan
N1 - DBLP License: DBLP's bibliographic metadata records provided through http://dblp.org/ are distributed under a Creative Commons CC0 1.0 Universal Public Domain Dedication. Although the bibliographic metadata records are provided consistent with CC0 1.0 Dedication, the content described by the metadata records is not. Content may be subject to copyright, rights of privacy, rights of publicity and other restrictions.
PY - 2022/1/1
Y1 - 2022/1/1
N2 - Cracks in oil paintings constitute an undesirable but unavoidable effect of time, deteriorating the painting quality. This work proposes a crack detection method that supports the physical restoration process of the artworks, providing a fissure map that allows the artist to visualize the pictorial layer and its flaws. This approach applies three image processing techniques to digitized oil paintings: oriented elongated filters, top-hat morphological filters and a K-SVD algorithm. Then, a post-processing stage based on K-Means is performed on the resulting binary maps to eliminate false positives. Finally, a pixel-by-pixel voting technique is applied to combine the binary maps. Our proposed framework has a better performance detecting craquelure when compared to other methods such as ADA Boost and convolutional neural networks. We obtained a recall of 0.8577, a probability of false alarm of 0.0779, a probability of false negatives of 0.1423, an accuracy of 0.7123, and an F1 value of 0.7783, which is amongst the best results for the state-of-the-art techniques.
AB - Cracks in oil paintings constitute an undesirable but unavoidable effect of time, deteriorating the painting quality. This work proposes a crack detection method that supports the physical restoration process of the artworks, providing a fissure map that allows the artist to visualize the pictorial layer and its flaws. This approach applies three image processing techniques to digitized oil paintings: oriented elongated filters, top-hat morphological filters and a K-SVD algorithm. Then, a post-processing stage based on K-Means is performed on the resulting binary maps to eliminate false positives. Finally, a pixel-by-pixel voting technique is applied to combine the binary maps. Our proposed framework has a better performance detecting craquelure when compared to other methods such as ADA Boost and convolutional neural networks. We obtained a recall of 0.8577, a probability of false alarm of 0.0779, a probability of false negatives of 0.1423, an accuracy of 0.7123, and an F1 value of 0.7783, which is amongst the best results for the state-of-the-art techniques.
UR - https://hdl.handle.net/20.500.12724/17547
UR - https://www.mendeley.com/catalogue/81cca638-9104-350e-8f00-44e2d978d183/
U2 - 10.1007/978-3-031-04447-2_22
DO - 10.1007/978-3-031-04447-2_22
M3 - Capítulo
AN - SCOPUS:85128957811
SN - 9783031044465
T3 - Communications in Computer and Information Science
SP - 329
EP - 339
BT - Communications in Computer and Information Science
A2 - Lossio-Ventura, Juan Antonio
A2 - Valverde-Rebaza, Jorge
A2 - Díaz, Eduardo
A2 - Muñante, Denisse
A2 - Gavidia-Calderon, Carlos
A2 - Valejo, Alan Demétrius
A2 - Alatrista-Salas, Hugo
PB - Springer Science and Business Media Deutschland GmbH
T2 - 8th Annual International Conference on Information Management and Big Data, SIMBig 2021
Y2 - 1 December 2021 through 3 December 2021
ER -