A Semiparametric Model for Hyperspectral Anomaly Detection

Table 3

Behavior of 𝛽 on different case studies.

	Simulated samples		Mean estimates	Variance estimates
Case studies	Parameters; 𝜇𝐴=2000;𝜎^2𝐴𝜇=200𝐵=1000;𝜎^2𝐵=100	𝛽(𝛼)	𝜇=∑^𝑛𝑖=1𝑡𝑖𝑔0(𝑡𝑖)𝜇2=(1/𝑛0)∑^𝑛0𝑖=1𝑦0𝑖𝜇3=(1/𝑛)∑^𝑛𝑖=1𝑡𝑖	𝑣^2=∑^𝑛𝑖=1𝑡^2𝑖𝑔0(𝑡𝑖)−𝜇^2𝑣^22=(1/𝑛0)∑^𝑛0𝑖=1(𝑦0𝑖−𝜇2)^2𝑣^23^=(1/𝑛)∑𝑛𝑖=1^(𝑡𝑖^−𝜇3^)2
Case 1	𝑦0∼i.i.d𝑁𝜇𝐴,𝜎^2𝐴𝑦1∼i.i.d𝑁𝜇𝐵,𝜎^2𝐵	−0.7500 (848.75)	1.9967𝑒+003; 1.9967𝑒+003; 1.4983𝑒+003	151.9466; 153.4815; 2.4982𝑒+005
Case 2	𝑦0∼i.i.d𝑁𝜇𝐴,𝜎^2𝐴i.i.d𝑁𝜇𝐵,𝜎^2𝐵𝑦1∼i.i.d𝑁𝜇𝐵,𝜎^2𝐵	−0.0073 (8.0110)	1.4990𝑒+003; 1.4990𝑒+003; 1.2494𝑒+003	2.4997𝑒+005; 2.5316𝑒+005; 1.8859𝑒+005
Case 3	𝑦0∼i.i.d𝑁𝜇𝐵,𝜎^2𝐵𝑦1∼i.i.d𝑁𝜇𝐵,𝜎^2𝐵	−0.0046 (4.5900)	999.8392; 999.8392; 999.6346	89.2284; 89.2574; 89.5693