Contrastive analysis of the performances of radar multi-look polarimetric detectors

Contrastive analysis of the performances of radar polarimetric detectors can provide important reference for its design, while the detection performance evaluation theory is still imperfect, especially lacking of the analysis results for multi-look case. Based on the complex Wishart distribution of the polarimetric radar multi-look covariance matrix, a contrastive analysis of the performance of 7 classical polarimetric detectors is presented using both Monte Carlo simulations and analytical method after an area under curve (AUC) index is proposed for evaluation. The ranking result for 4-look case is:optimal polarimetric detector (OPD) > polarimetric whitening filter (PWF)≈identity likelihood ratio test (ILRT) > optimal sapn detector (OSD) > polarimetric matched filter (PMF)≈ span detector (SD) > single channel detector (SCD). At the same time, a novel approach to compute the constant false-alarm rate (CFAR) threshold is proposed based on Gamma function, and two kinds of general mathematical expressions of the theoretical performances of those detectors with linearly weighted form are also derived for multi-look case. Finally, the polarimetric SAR data is used to verify the above results.