Many instrumentation and early warning/BMD radars perform object classification based on radar signature measurements, for example, sorting reentry vehicle (RV) vs. decoy. This is usually obtained through deceleration of the body by the atmosphere, wake effects (mean and spread), micro dynamic motion (nose tip precession), polarization, range profile, inverse synthetic aperture radar (ISAR) imaging, radar cross-section (rcs) statistics, etc.

Typical estimators include Bayesian approaches. The K factor describes the ability to resolve two types of objects, that is, the separation of their probability density functions normalized to the spread of the density function.

Many lightweight traffic decoys (e.g., balloons) can be placed on a post boost vehicle (PBV) by replacing an RV, but the ability of the lightweight decoy to penetrate the defense is less than that of a heavier replica decoy. Munkers algorithm can be used for optimally assigning objects seen on one sensor to those seen by another sensor, that is, handover or target object mapping (TOM).

Much work has been performed in the past in identifying or classifying battlefield vehicles (e.g., truck, jeep, tank) based on high-range resolution measurements. A priori measured range profiles at various angles can be stored to be matched against by an unknown object.

Sometimes features are extracted from the data such as spacing between largest spikes, order of magnitude of spikes, etc. Some of the work has involved the use of neural networks.

The simplest imaging radars use high range resolution with Doppler processing, that is, FFTs within the range cells. For a rotating object, the Doppler frequency increases with distance from the axis of rotation and hence, maps cross range intoDoppler to produce two-dimensional ISAR images.

Range walk will limit Doppler resolution since it determines how many pulses can be processed in the Doppler filter. The crossrange resolution is related to the angle through which the target rotates during the coherent processing.

ISAR is similar to the conventional noncoherent tomography (radon transform, back projection) used in X-ray processing. Since it is only the relative motion between radar and target that is important, the turning object in ISAR is equivalent to a stationary target and a synthetic circular SAR, that is, aircraft flying a circle about the target.

More advanced ISAR imaging radars use polar processing to avoid the range walk problem. The most advanced imaging radars use extended coherent processing where an image is created by coherently overlaying images for several complete rotations of the object. Maximum entropy method(MEM) techniques can be used to extend the bandwidth to provide sharper images for a given actual RF bandwidth.

Airborne synthetic imaging radars (i.e., conventional SAR) use a small aperture on a moving platform. By storing the pulses and coherently combining them, a large synthetic array can be constructed that is focused at all ranges.

The effective synthetic pattern is actually a 2-way pattern and the cross-range resolution at every range is about the same as the size of the physical antenna on the aircraft. At each range, the phases from a scatterer produces a quadratic runout (i.e., LFM) that varies with range.

Each range cell is match filtered yielding a pulse compression in the azimuth direction. Since Doppler frequency is mapping into cross range, moving objects such as a train create a range-Doppler coupling and may image off the tracks.

Stereoscopic imaging can be performed by using SAR mapping from two aircraft ormultiple displaced apertures on the same aircraft. The phase difference in a common pixel for the two apertures will provide height data within the pixel.

No comments:

Post a Comment

Related Posts Plugin for WordPress, Blogger...