Robust CFAR Detector Based on Censored Harmonic Averaging in Heterogeneous Clutter

In this article, a robust constant false alarm rate detector (CFARD) is proposed for dense outlier situations such as the presence of high-density interfering targets within the reference window. Harmonic mean mitigates the impact of large outliers better than the arithmetic mean. We propose a CFARD called censored harmonic averaging (CHA)-CFARD, by using the modified harmonic averaging instead of the common arithmetic averaging in order to estimate the local clutter level. In practice, the number of cells which contain outliers is not known precisely. This poses a major challenge for censoring and truncating methods which aim to eliminate outliers from the reference window. The proposed detector reduces the undesirable effects of strong outliers without this restriction. Simulation results show that the CHA-CFARD outperforms the well-known CFARDs in the presence of interfering targets. Moreover, at the clutter edges, the proposed method shows an acceptable performance. Experimental results with TerraSAR-X images demonstrate that the proposed method is also effective and efficient for ship detection applications in nonhomogeneous sea clutter environments.