Abstract:

A high resolution range profile recognition method is proposed based on the global scattering center model. The model is extracted offline using wideband data measured on a sparse spatial grid. Then the model is projected and compared with the 1D scattering center feature extracted from the range profiles to achieve  target recognition.In extracting the global scattering center model, the stable scattering center is defined and the 1D projections on the sparse spatial grid are associated based on their stability. In range profile recognition, a proper matching function is designed to deal with the number, position and amplitude mismatch between the 1D scattering centers extracted from the range profile and the projected scattering center model. Experiments show that the recognition accuracy of the modelbased method is as good as the template-based method for well established model. Moreover, the model-based method outperforms the templatebased method in storage consumption and flexibility.