Computational protocol: A New Quantitative Method for the Non-Invasive Documentation of Morphological Damage in Paintings Using RTI Surface Normals

Similar protocols

Protocol publication

[…] In this monitoring method the image alignment between imaging datasets is critical because mis-aligned pixels between imaging sessions would show up as topographical changes. Since it is not possible to repeatedly replace and image an object on the spatial scale of a pixel, rigorous digital alignment methods were employed.The repeatability of the method was measured by recording the data in five separate imaging sessions: for every imaging session the camera and the object were repositioned in the same position. Then, the open source software RTI Builder was used to build the RTI images from the raw image set and the normals were extracted as described by Macdonald et al., and Gautron et al. [,], using both Python version 2.7.3 and Matlab R2010a (the MathWorks, Natick, MA, USA).The extracted normals of each replicate can be represented as three bi-dimensional images (one for each Cartesian coordinates x, y, z). In order to elaborate and compare the object over time, as described later in the paper, the images of x, y, and z must be aligned between each other. An open source image processing package, a version of ImageJ, called FIJI was used [,] to match the coordinate images of the five replicates. The color images, captured to build the RTI files, were employed to calculate an elastic transformation function (SIFT and bunwarpj tools were used) that was then used to align the x, y, and z coordinate images. In this way it was possible to align the images collected during different imaging sessions without positioning the painting in the same place. Moreover, using the elastic transformation function, which was based only on the color images and on their features, the method was able to align the coordinates images taken after a damage: in this case, the alignment has been performed without influencing the damaged areas because the information related to the normals is not accounted by the raw color images used to obtain the elastic function. […]

Pipeline specifications

Software tools ImageJ, bUnwarpJ
Application Microscopic phenotype analysis
Organisms Homo sapiens
Diseases Nervous System Diseases