IBISA: Making Image-Based Identification of Ancient Coins Robust to Lighting Conditions

Abstract

The IBISA (Image-Based Identification/Search for Archaeology) system manages databases of digital images of archaeological objects, e.g. ancient coins, and allows the user to perform searches by examples. IBISA was de- signed to help the user decide, from their images, if two objects (coins) are either the same, come from the same matrix (die), share resemblance in style, or are completely different. The system searches for similarities in the databases using a registration method that must be resilient to the viewing conditions. Based on the Fourier trans- form, it cancels rigid transforms among images. Sub-pixel accuracy can be achieved with a very simple technique. However lighting conditions remain an issue. Fortunately, it is possible to extend this registration method to a light-independent model, considering the elevation or normal maps instead of intensity. The model is also useful for interactive visualization and museography. Although this model registration is now resilient to all viewing conditions, it is not practical in real scenarios where the target is a single image, from which a model can hardly be derived. Finally, a hybrid approach is investigated, with a target image but a model of the reference. It is more realistic, resilient to light conditions, gives excellent results with translations, but shows limitations for rotations.