Image processing methods (IPM) can be helpful for the restoration of old artworks. One of the possibilities how IPM can be employed is virtual restoration. Having information from various sources (visual images, UV and IR images, old photographs of the artwork, chemical point analyzes) we can create a model which would be closer to the artwork at the time of its creation than the current state is. Information from all data sources can be combined and fused, structures can be extrapolated and "improved". All artificial interventions are logged and possibly marked in the resulting model, to make sure the historical value is not biased.
We worked with four data sources - images from regular camera, old photographs from the 60's, and images taken in wide and narrow ultraviolet bands, as shown below.
All four data channels are geometrically alligned and then fused together, to have all important details visible in one putput image. Part of the painting we are interested in are then more easily detectable, as in the case of lacunas.
The data from different channels can be compared (see Figure below, where detected lacuna border from one data channel is wider than in the case of an other acquisition bandwidth. This is caused by the diferent visibility of materials in different wavelengths. )
The processed and fused data can be used for the adaptive information extrapolation. Knowing the exact chemical content at certain position, this information can be expanded till a difference between processed and the known position is not over given limit.
|Contact person:||Barbara Zitova|
|Involved people:||Jan Blažek, MSc. student|
|Involved extern:||J. Hradilová, AVU, ALMA|