Polarimetric ATI slow target detection in a log likelihood framework

Stacy, N.J.S.; Preiss, M.

doi:10.1109/igarss.2013.6723498

Cited by 5 publications

(13 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Under the binary hypothesis of

H_{1}

and

H_{0}

, that is, target exists or not, the likelihood ratio test of the full polarimetric interferometric covariance matrix has the following form

L = \prod_{i = 1}^{N} \frac{p ({bold-italick}_{i} | {bold-italicC}_{c})}{p ({bold-italick}_{i} | {bold-italicC}_{t})} ≷_{H_{0}}^{H_{1}} T_{LRT}

where

T_{LRT}

is the detection threshold. The detection metric can then be rewritten as [16]

D_{LRT} = trace (][, ()(, C_{c}^{- 1} - C_{t}^{- 1}) \sum_{i = 1}^{N} {bold-italick}_{i} {bold-italick}_{i}^{normalH})

where the subscript c and t denote clutter and moving target,

p false(k_{i} falsefalse| C_{c} false)

and

p false(k_{i} falsefalse| C_{t} false)

are the prior probability under the condition that clutter and moving target exist, respectively, and N is the number of freedom [22] which is also known as the nominal Equivalent Number of Looks (ENL) [23]. A high similarity of the mathematical formula could be found when

D_{LRT}

Section: Slow Targets Detectors Of At‐polinsarmentioning

confidence: 99%

“…A high similarity of the mathematical formula could be found when

D_{LRT}

is compared with the Optimal Polarimetric Detector (OPD) [20], thus, the exact probability of detection or false alarm can be derived using the same method of characteristic function in reference [22], also both of them can achieve the optimal performance of target detection. The optimal detection performance of AT‐POLINSAR GMTI has been presented as ROC curves in [16] which has also offered a reference in the design of new GMTI detectors.…”

Section: Slow Targets Detectors Of At‐polinsarmentioning

confidence: 99%

“…The optimal detector obtained by LRT in [16] provides the upper bound of the detection performance in AT‐POLINSAR GMTI, but LRT is difficult to implement in reality since it requires the prior information of the target and clutter. As the analysis points out in the previous section, LRT detects the difference between the two covariance matrices by computing the Euclidean distance, the longer the distance is, the easier it is to detect.…”

Section: Slow Targets Detectors Of At‐polinsarmentioning

confidence: 99%

“…So far there are two methods that can be used for AT‐POLINSAR GMTI: the first is based on the optimal interferometry in references [9, 15], which can locate and image moving targets in foliage by building an optimal polarisation from full polarimetric channels [1], but this optimal model oriented by the highest interferometric coherency, which is originally built for altitude measurement of POLINSAR, might not achieve the best detection probability. The second is to use the log‐likelihood‐ratio test (LRT) of covariance matrix formulated by vector interferometry [16]. Although these two approaches can be applied to AT‐POLINSAR in the future, both of them rely on the prior information of target and also lack statistical analysis of the detection metrics.…”

Section: Introductionmentioning

confidence: 99%

“…(ii) how to build novel metrics that do not require any prior knowledge? Inspired by POLSAR detectors, appropriate polarimetric ATI statistical model and dimension reduction make vital contribution to solving such problems [16].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations