Tightly based on"Project 973", target recognition problems basing on High Resolution Radar Profile (HRRP) are discussed in the thesis, and two novel approaches of target recognition are proposed.After analyzing the theory of target scattering and characters of HRRP, the approach of target recognition is presented, which is based on the statistical features of HRRP, including peak value , scale, radial energy accumulation variance , and FFT-MDMT characters. Then D-S evidence theory is demonstrated, which gives an improvement in fusing the recognition results of separate features. And the validity of the approach is verified by the simulation tests of three ships target recognition problem.A novel theory—RST is introduced, and is applied in target recognition of HRRP. The algorithms of attribute reduction are studied and compared. Then, an improved heuristic algorithm of attribute reduction of RST is presented, which can eliminate the redundancy. Regarding the characters of RST, the preprocessing of HRRP is designed, including data partitioning, normalization, binary multi-class quantification. Finally, the target recognition algorithm is designed and is simulated on the Matlab platform. |