Number of targets detection method with FCM‐based granulation–degranulation mechanism

Abstract

In this letter, the authors elaborate on a novel method to detect the number of targets in low SNR and small snapshots through using the Fuzzy C-Means clustering based granulation-degranulation mechanism. In the developed scheme, the eigenvalues of the array output correlation matrix are regarded as a time series, and granulated into a pair of ordered information granules. The information about the number of targets is included in the information granules. Subsequently, the degranulation mechanism is used to modify the cost function of the granulation mechanism to improve the quality of the information granules, which can increase the detection accuracy of the number of targets. Finally, the number of targets is determined by the prototype matrix and partition matrix generated by the enhanced version of the granulation. Explicit analysis and derivation of the proposed scheme are presented.