Reverse-Engineering techniques are commonly used to generate or update the CAD model of a single physical object. However, the reverse engineering of a whole assembly is still very tedious and time consuming.This is mainly due to the fact that the complete definition of the final digital mock-up relies on the integration of multiple sources of heterogeneous data, such as point clouds, images, schemes or any type of digital representations which are not yet fully supported by actual software. This paper proposes a new method and tool to better integrate those multi-representations and to speed up the reconstruction process which could therefore become adapted to the reconstruction of large mechanical assemblies such as in automotive field. The conceptual model of the methodology suggested enables to extract geometrical mark from the heterogeneous data thanks to segmentation and to identify mechanical components. In our approach,“signature” plays a key role in the identification and it is considered as a set of characteristics to describe an object. This article presents a demonstrator to illustrate this methodology using an example from automotive domain.