Table 3 Performance comparison, single target (S) and multiple close targets (M) in noise, \(P=6\)
From: A neural network framework for binary classification of radar detections
Detector (trained on) | S: \(P_{\mathrm{D}}\) | S: \(P_{\mathrm{FA}}\) | M: \(P_{\mathrm{D}}\) | M: \(P_{\mathrm{FA}}\) |
|---|---|---|---|---|
NN (N) O1, 50x1 | 0.76 | \(5.78\times 10^{-8}\) | 0.34 | \(3.92\times 10^{-8}\) |
NN (N) O2, 50x1 | 0.80 | \(5.78\times 10^{-8}\) | 0.51 | \(1.96\times 10^{-8}\) |
NN (N) O3, 50x1 | 0.82 | \(9.64\times 10^{-9}\) | 0.78 | 0 |
NN (N) O1, 50x2 | 0.76 | \(7.52\times 10^{-7}\) | 0.64 | \(4.61\times 10^{-7}\) |
NN (N) O2, 50x2 | 0.81 | \(8.67\times 10^{-7}\) | 0.77 | \(3.92\times 10^{-7}\) |
NN (N) O3, 50x2 | 0.82 | \(3.95\times 10^{-7}\) | 0.80 | \(1.07\times 10^{-7}\) |
NN (N+C) O1, 50x1 | 0.70 | 0 | 0.41 | \(9.81\times 10^{-9}\) |
NN (N+C) O2, 50x1 | 0.79 | 0 | 0.63 | 0 |
NN (N+C) O3, 50x1 | 0.79 | \(1.92\times 10^{-8}\) | 0.70 | \(9.64\times 10^{-9}\) |
NN (N+C) O1, 50x2 | 0.72 | \(2.89\times 10^{-8}\) | 0.60 | \(2.35\times 10^{-7}\) |
NN (N+C) O2, 50x2 | 0.79 | \(2.89\times 10^{-8}\) | 0.75 | \(2.02\times 10^{-7}\) |
NN (N+C) O3, 50x2 | 0.82 | \(5.01\times 10^{-7}\) | 0.79 | \(5.11\times 10^{-7}\) |
CMSO-CFAR | 0.82 | \(2.43\times 10^{-3}\) | 0.82 | \(2.38\times 10^{-3}\) |
SO-CFAR | 0.77 | \(1.19\times 10^{-5}\) | 0.62 | \(1.10\times 10^{-5}\) |
GO-CFAR | 0.74 | \(3.85\times 10^{-8}\) | 0.09 | \(1.96\times 10^{-8}\) |