Improved revealing of hidden structures and defects for historic art sculptures using poisson image editing

Abstract

[EN] Radiography is a non-destructive tool and offers the acquisition of detailed information on the internal features of sculptures as a cultural heritage. However, radiographs contain different levels of blurriness mainly caused by the detection of scattered X-rays. Reduction of image blurriness provides improved contrast in targeted areas which enhances the extraction of information from the selected regions and features of the radiographs. In this study, we applied a set of convolution methods to a group of radiographic images of historic sculptures. Radiographs of the objects were provided with the associated documentation from the collection of the Radiographic Inspection Laboratory of the Universitat Politecnica de Valencia. The selection of the particular objects was based on the difference in the materials used in their construction i.e. the objects were made of wood, paper, or wax. The Poisson Image Editing (PIE) based on L-2-norm was applied for image enhancement of digital radiography images. The results showed that the PIE method was effective in selective region enhancement of the radiographic image contrast enabling better visualization of the objects' internal structures. The application of the implemented algorithm enabled the conservators and radiographers involved in the study to improve the visualization of the sculptures' internal features and defects enhance the defects' evaluation.