A Decision Aiding System Based on the Choquet Integral Applied to Quality Evaluation of Composite Material Parts

The purpose of this paper is to develop a decision aiding system for quality evaluation of composite material parts. The application deals with the detection of different regions inside parts. X-ray images of composite material parts are classified by means of three different methods: Haar wavelets, Gabor wavelets and PCA of gradient vectors. The obtained attributes are transformed into similarity maps to be expressed in the same space. Moreover, this representation is more understandable by experts. These similarity maps are then aggregated to obtain cartography of the part. The aggregation process is based on the 2-additive Choquet integral to take redundancy, complementarity and independence of similarity maps into account. A learning algorithm is also proposed for the Choquet integral parameter estimation based on reference regions pointed by experts. The experiments show the system performance