Fusion de données gammagraphiques et ultrasonores

This work deals with the three dimensional reconstruction from Xray and ultrasonic images. Such an issue belongs to the field of data fusion since the data provide complementary information. The two sets of data are independently related to two sets of parameters: Xray attenuation and ultrasonic reflectivity. The fusion problem is adressed in a Bayesian framework; the kingpin of the task is then to define a joint prior model for both attenuation and reflectivity. Dealing with the joint prior model, we propose to correlate the derivative of the attenuation and the reflectivity. Processing examples demonstrate the validity of the fusion approach and the proposed method is robust towards mismatching of the two sets of data.