Event detection in multisource imaging: application to fine arts painting analysis

The standard statistical tools proposed in the literature for revealing changes between frames in a temporal sequence of images are ill suited to the detection of significant differences or changes (events) between pictures of a given scene taken at different wavelengths or with different sensors. A method for modeling a pair of multisource images so as to reveal significant changes between the two images is described. The method involves models which describe each image separately, plus an interimage model describing the relationship between them. The use of the models in detecting changes is described in detail and illustrated by an application to the fine arts, that of finding an underpainting from a visible/X-ray pair of images of the same painting.<
>