Abstract:
In order to enhance radar target recognition performance of high-resolution range profile (HRRP) under low signal-to-noise ratio (SNR), a novel HRRP sequential recognitio...Show MoreMetadata
Abstract:
In order to enhance radar target recognition performance of high-resolution range profile (HRRP) under low signal-to-noise ratio (SNR), a novel HRRP sequential recognition method utilizing the correlation among scattering centers is proposed in this letter. In this method, the correlative information among the prominent scattering centers is considered in the covariance matrix and the sequential recognition is carried out through the long short-term memory (LSTM) network to fully excavate the information of HRRPs. Moreover, we introduce a noise-robust recognition algorithm to renew the Gaussian-trained model by the estimated variance of the noise. Experimental results indicate that the proposed method can acquire higher recognition rates and better robustness by introducing correlative information and sequential processing.
Published in: IEEE Geoscience and Remote Sensing Letters ( Volume: 20)