Paint loss detection in old paintings by sparse representation classification

Shaoguang Huang, Wenzhi Liao, Hongyan Zhang, Aleksandra Pizurica

Research output: Contribution to conferencePaper

Abstract

In this paper, we explore the potential of sparse representation classification (SRC) in digital painting analysis, with the aim to aid the conservation/restoration treatment of old masterpieces. The focus is on detecting paint losses using multimodal acquisitions (such as optical images taken at different time instances before and during the conservation treatment, infrared images and digital radiography images). While SRC has been applied before in different scenarios, the present application requires some specific adaptations due to the nature and the size of the data, as well as the uncertainty to the labelled samples. Our initial results are very promising, compared to some more traditional or commonly used classification approaches, such as linear regression classification and support vector machines.
Original languageEnglish
Pages62-64
Number of pages3
Publication statusPublished - 2016
EventiTWIST 2016: international - Traveling Workshop on Interactions between Sparse models and Technology - Aalborg, Denmark
Duration: 24 Aug 201626 Aug 2016

Workshop

WorkshopiTWIST 2016: international - Traveling Workshop on Interactions between Sparse models and Technology
Abbreviated titleiTWIST2016
CountryDenmark
CityAalborg
Period24/08/1626/08/16

Keywords

  • paint loss detection
  • sparse representation classification (SRC)
  • image classification

Cite this

Huang, S., Liao, W., Zhang, H., & Pizurica, A. (2016). Paint loss detection in old paintings by sparse representation classification. 62-64. Paper presented at iTWIST 2016: international - Traveling Workshop on Interactions between Sparse models and Technology, Aalborg, Denmark.