Sensor Fusion for Quality Control of Agricultural Products

Abstract

Different tools are described in order to check the complementarity or the redundancy of different sensors before applying a fusion process. Different techniques are described for applying the fusion process at different levels (decision and feature level). Different tools are described in order to check whether the fusion process does or does not improve the performance of the decision. These techniques are successfully applied to two different kinds of fruits: peaches for firmness assessment, and melons for quality prediction, though some problems remain with the inconsistency of the classification provided by an expert in the case of melons.