Temporal appearance change detection using multi-view image acquisition

Appearance change detection is a very important task for applications monitoring the degradation process of a surface. This is especially true in Cultural Heritage (CH), where the main goal is to control the preservation condition of an artifact. We propose an automatic solution based on the estimation of an explicit parametric reflectance model that can help the user in the detection of the regions that are affected by appearance changes. The idea is to acquire multi-view photo datasets at different times and to compute the 3D model and the Surface Light Field (SLF) of the object for each acquisition. Then, we compare the SLF in the time using a weighting scheme, which takes account of small lighting variations and small misalignments. The obtained results give several cues on the changed areas. In addition, we believe that these can be used as good starting point for further investigations.