T. Bucciarelli

Adaptive polarimetric target detection with coherent radar. II. Detection against non-Gaussian background

The derivation of a completely adaptive polarimetric coherent scheme to detect a radar target against a Gaussian background is presented. A previously proposed Generalized Likelihood Ratio Test (GLRT) polarimetric detector is extended to the case of a general number of channels; this exploits the polarimetric characteristics of the received radar echoes to improve the detection performance. Together with the fully adaptive scheme, a model-based detector is derived that has a lower estimation loss. A complete theoretical expression is derived for the detection performance of both proposed polarimetric detectors. They are shown to have Constant False Alarm Rate (CFAR) when operating against Gaussian clutter, but to be sensitive to deviations from the Gaussian statistic. The application to recorded radar data demonstrates the performance improvement achievable in practice.

Adaptive polarimetric target detection with coherent radar

A CFAR adaptive polarimetric generalized likelihood ratio test (GLRT) detector is proposed for the coherent detection of radar targets against a Gaussian background and its performance is fully characterized. A model based version of such a detector is also derived by applying a similar GLRT approach to a structured covariance matrix. This latter detector is shown to reduce the adaptivity losses and thus to reduce the homogeneous region, which is required to estimate the clutter covariance matrix. The application to live radar data demonstrates the performance improvement achievable in practice by exploiting the polarimetric information. In particular adding the HV channel to the two co-polarized channels can provide a sensible performance increase with respect to the HH and VV only. This is especially true for man-made targets having cross-polarized response higher than the clutter. When the target has a lower cross-polarized return than the clutter, a much lower improvement is available but there is not a sensible adaptivity loss and the CFAR characteristic is always enforced against the Gaussian background.

PBR activity at INFOCOM: Adaptive processing techniques and experimental results

The paper describes the results achieved at the INFOCOM Department of the University of Rome ldquoLa Sapienzardquo on passive bistatic radar (PBR). The developed adaptive signal processing techniques are introduced for: improved disturbance cancellation, multipath removal, and multifrequency target detection. The development of experimental wide-band PBR prototypes is described, as opposed to single frequency channel PBR systems, giving special attention to the system design guidelines and test procedures. The experimental results achieved using the wide-band PBR prototypes are shown, there including the demonstration of effectiveness of the techniques above. The results are shown to be largely consistent with live air traffic control registrations.

GPR for archaeological investigations: real performance assessment for different surface and subsurface conditions

This paper assesses GPR (Ground Penetrating Radar) performance for archaeological investigations. We exploit extensive data acquisition campaigns to characterize the instrument behavior in different operative conditions and we select ad hoc processing techniques to optimally extract target signatures from the collected data. The extensive analysis demonstrates the real effectiveness of GPR surveys in archaeological investigations.

CFAR coherent radar detection against K-distributed clutter plus thermal noise

The paper shows the non-CFAR characteristic of the asymptotically optimum detector (AOD), when applied to a mixture of K-distributed clutter plus thermal noise, as the target Doppler frequency is varied. The upper bound to the performance against such composite disturbance is investigated by considering an ideal-AOD detector, which assumes a knowledge of the local clutter power in the cell under test. A practical scheme is obtained by replacing the known local power value with its estimate. The resulting detection scheme, named locally adaptive-AOD (LA-AOD) shows a proper CFAR characteristic, but yields strong detection losses in the same Doppler frequency region where the original AOD operates effectively with CFAR. Finally, a combined detector, Comb-AOD, is proposed, which uses the benefits of both the AOD and LA-AOD structures, and its detection performance is fully characterized as the clutter spikiness and correlation vary.

Statistical analysis of multipolarisation/multifrequency SAR images of the sea surface

A statistical analysis of SAR images of the sea surface is presented. The analysis shows a good agreement between compound-Gaussian distributions and statistical properties of the sea radar echoes. A relation between probability density function (pdf) of local backscattered power and local incidence angle is derived. This is used to simulate SAR images and to make inferences on the sea state.

Performance assessment of wide-band pulsed GPR for archaeological applications

The paper describes the results of an experimental application of a pulsed, wide band ground penetrating radar (GPR) for archaeological applications. The experiment was used to validate the detection capability of commercial GPR and evaluate the performance increase available by using ad-hoc signal processing techniques. The comparison with excavation allows the authors to assess the reliability of the radar sounding.

Rank detectors in correlated K-distributed clutter

The paper deals with the performances of the first and second order rank detectors in presence of correlated K-distributed clutter. First, assuming a perfect knowledge of both clutter shape factor and texture correlation, the detection probability of the rankers is compared to the ideal performance and to that of a cell-averaging detector. Then, an on-line estimator of the shape factor in the correlated environment is introduced, and the corresponding detection losses are analyzed.

Estimating radar clutter distributions via neural networks

A new approach to the solution of constant false alarm rate (CFAR) problems in complex, unknown environment is proposed. An adaptive system is described in which a neural network properly transforms the input random process which describes the clutter amplitude. The network proposed is a modified neural network (Tchebychev neural network), specifically designed for the problem at hand, whose neurons implement Tchebychev polynomials up to a proper order. Experimental results are presented referring to Weibull and Rayleigh input distributions.