Research on the Micro-Motion Patterns Recognition Method Based on Characteristic Value Entropy

The echo of target with micro-motion contains the electromagnetic scattering characteristics and movement characteristics, which plays important roles in the target identification. Precession period is one of the important characteristics on the analysis of the micro-motion characteristics. Aimed at the phenomenon of extracting the cycle of no precession target, in the thesis a method is proposed by combing period extraction with eigenvalue entropy. The original echo is constructed by establishing autocorrelation matrix and the eigenvalues are obtained through eigenvalue decomposition. Then the period is extracted after calculating the eigenvalue entropy. The method can eliminates some targets with no obvious precession and minishes the error of period extraction, which lays the foundation for the recognition of the final target class.