Abstract: Inspection of industrial products entails the “Model Based Design” paradigm with an apparatus associated with a digital clone of the real acquisition system. The digital clone reproduces in a virtual environment the features of said system and of the product affected by anomalies to be inspected. By generating virtual images of the product through simulations on said digital clone and processing virtual images through a vision system including a novel data fusion module, it is possible to define the optimal architecture and parameterization of the real acquisition system that maximize the likelihood of identifying anomalies in products even in conditions that are difficult or expensive to replicate in the real world. This reduces time in designing and set-up as well as costs of an inspection apparatus. Furthermore it enables to reconfigure the apparatus to products and/or anomalies other than those for which it was initially designed.