Since last decades, microwaves have received tremendous attention as an interesting tool for material characterization. In general, standard microwave measurement methods require cutting and polishing of samples to put it in a suitable waveguide or cavity. However, several methods have been developed in order to permit a non-destructive measurement. A well-known method is based on coaxial open-ended waveguide, which is used as a sensor for dielectric characterizations. Moreover, with the requirement of new forms, developing mathematical model for each one is not convenient. Indeed, the complex structures required in the industrial field can be perfectly designed with high-performance three-dimensional software. Many attempts have been done to solve the conversion problem by proposing different algorithms. Nevertheless, they are sensitive for complex structure that contains transition part. In this paper, we propose a dielectric measurement method based on the use of coaxial waveguide. A novel algorithm for dielectric characterization of complex structures is also presented, which is based on the joint use of artificial neuronal networks and finite element method. The proposed algorithm aims to find the dielectric characterization for complex structures. Experimental evaluations applied to solid and liquid dielectrics confirm the validation of the proposed algorithm.