Raytheon Company

Bae Systems Information and Electronic Systems Integration Inc.

Anthony M Sommese

Ian S Robinson

Bradley A Flanders

Daniel Engheben

Valeri I Karlov

John D Hulsmann

Aaron Maestas

Christopher J Cormier

Owen Lewis

The system and method for super resolution processing at long standoff distances in real-time. The system collects a series of image frames and estimated the shift, rotation, and zoom parameters between each of the image frames. A matrix is generated and then an inversion is applied to the matrix to produce a super resolution image of an area of interest while mitigating the effect of any bad pixels on image quality. In some cases, the area of interest is user-defined and in some cases image chips are provided by tracking software. A fast steering mirror can be used to steer and/or dither the focal plane array.

Real-time super resolution at long standoff ranges

A system and method for motion compensation in translational and rotational image smear using a two-axis fast steering mirror and a single-axis gimbal which are controlled simultaneously. A pointing and control module calculates an objective function which can be optimized to find an optimized de-roll rate to compensate for camera boresight rotation in real-time.

Optimized motion compensation via fast steering mirror and roll axis gimbal

Technology for determining a position of a platform is described. A location of a horizon line can be determined using a sensor onboard the platform. One or more celestial objects in the sky can be detected using the sensor onboard the platform. Differential angular measurements between the horizon line and at least one of the celestial objects in the sky can be determined over a duration of time. The position of the platform can be determined based on the differential angular measurements between the horizon line and the celestial objects.

Synthetic digital sextant for navigation

Methods and systems are provided for estimating background spectral content in a hyperspectral imaging (HSI) scene. A HSI processor computes a scene covariance matrix for each of a plurality of sparsely sampled pixel sets, identifies and removes the spectral content of contaminating pixels from the covariance matrices, and checks the consistency among the plurality of decontaminated covariance matrices, iteratively re-sampling and re-computing said matrices until an acceptable consistency is achieved, and then computes a final decontaminated covariance matrix representative of the background spectral content of the scene. Alternate approaches to pixel sampling, and/or using fewer spectral dimensions than are available for the pixels are presented.

Independent covariance estimation and decontamination

A rapid anomaly detection approach with corresponding method and system to detect anomalies in scene pixels making up a hyperspectral scene, efficiently, is presented. The approach includes tailoring an approximation of an anomaly score for each scene pixel, individually, based on an 'intermediate anomaly score.' The intermediate score is computed using a portion of the terms used to compute the anomaly score. Scene pixels with low intermediate anomaly scores are removed from further processing. The remaining scene pixels are further processed, including computing anomaly scores to detect anomalies in these pixels. Advantageously, examples of the RAND approach process a few terms of all scene pixels, eliminate most scene pixels, and calculate more terms on high anomaly scoring scene pixels as needed.

Rapid anomaly detection (RAND)

A rapid target detection approach with corresponding method and system to detect targets in scene pixels, efficiently, is presented. The approach includes tailoring an approximation of a target score for each scene pixel, individually, based on an 'intermediate target score.' The intermediate target score includes a portion of the terms used to compute the target score. The portion is selected by computing a signal-to-clutter ratio (SCR) for a spectral reference associated with a target and ranking the terms by their contribution to the SCR. Scene pixels with low intermediate target scores are removed from further processing. The remaining scene pixels are further processed, including computing target scores to detect targets in these scene pixel. Advantageously, examples of the approach process a few terms of all scene pixels, eliminate most scene pixels, and calculate more terms on high target scoring scene pixels as needed.

Rapid iterative detection (RID)

The disclosure provides a filtering engine for selecting a subset of hyperspectral imaging wavebands having information useful for detecting a target in a scene. Selecting these wavebands, called 'sparse bands,' is an iterative process. One or more search techniques of varying computational complexity are used in the process. The techniques rely on various selection criteria, including a signal to clutter ratio that measures the 'goodness' of band selection. A convenient example of the filtering engine uses several of the techniques together in a layered approach. In this novel approach, simpler computational techniques are applied, initially, to reduce a number of bands. More computationally intensive techniques then search the reduced band space. Accordingly, the filtering engine efficiently selects a set of sparse bands tailored for each target and each scene, and maintains some of the detection capability provided with a full set of wavebands.

Sparse adaptive filter

Provided are examples of a detecting engine for determining in which pixels in a hyperspectral scene are materials of interest or targets present. A collection of spectral references, typically five to a few hundred, is used in look a through a million or more pixels per scene to identify detections. An example of the detecting engine identifies detections by calculating a kernel vector for each spectral reference in the collection. This calculation is quicker than the conventional Matched Filter kernel calculation which computes a kernel for each scene pixel. Another example of the detecting engine selects pixels with high detection filter scores and calculates coherence scores for these pixels. This calculation is more efficient than the conventional Adaptive Cosine/Coherence Estimator calculation that calculates a score for each scene pixel, most of which do not provide a detection.

Kernel with iterative computation

Provided are examples of a detecting engine for identifying detections in compressed scene pixels. For a given compressed scene pixel having a set of M basis vector coefficients, set of N basis vectors, and code linking the M basis vector coefficients to the N basis vectors, the detecting engine reduces a spectral reference (S) to an N-dimensional spectral reference (SN) based on the set of N basis vectors. The detecting engine computes an N-dimensional spectral reference detection filter (SN*) from SN and the inverse of an N-dimensional scene covariance (CN). The detecting engine forms an M-dimensional spectral reference detection filter (SM*) from SN* based on the compression code and computes a detection filter score based on SM*. The detecting engine compares the score to a threshold and determines, based on the comparison, whether the material of interest is present in the given compressed scene pixel and is a detection.

Post compression detection (PoCoDe)

The disclosure provides a filtering engine for selecting sparse filter components used to detect a material of interest (or specific target) in a hyperspectral imaging scene and applying the sparse filter to a plurality of pixels in the scene. The filtering engine transforms a spectral reference representing the material of interest to principal components space using the eigenvectors of the scene. It then ranks sparse filter components based on each transformed component of the spectral reference. The filtering engine selects sparse filter components based on their ranks. The filtering engine performs the subset selection quickly because the computations are minimized; it processes only the spectral reference vector and covariance matrix of the scene to do the subset selection rather than process a plurality of pixels in the scene, as is typically done. The spectral filter scores for the plurality of pixels are calculated efficiently using the sparse filter.

Sparse reduced (spare) filter

A method for reducing dimensionality of hyperspectral images may include receiving a hyperspectral image having a plurality of pixels. A basis vector set including a number of members may then be established, wherein each of the members comprises a basis vector. For each of the plurality of pixels, a spectral vector for the pixel may be read and decomposed with the members of the basis vector set to derive a residual vector for the pixel. A basis vector for the pixel may then be added to the members of the basis vector set if the residual vector for the pixel has a magnitude exceeding a predetermined threshold, and the basis vector set may then be optimized to eliminate one of the members of the basis vector set, whereby the optimized basis vector set includes the number of members. A system configured to perform the method may also be provided.

System and method for reducing dimensionality of hyperspectral images

A hyperspectral imaging sensor and an adaptive spatial spectral processing filter capable of detecting, identifying, and/or classifying targets having a spatial extent of one pixel or less includes a sensor that may be oversampled such that a pixel is spatially smaller than the optical blur or point spread function of the sensor. Adaptive spatial spectral processing may be performed on hyperspectral image data to detect targets having spectral features that are known a priori, and/or that are anomalous compared to nearby pixels. Further, the adaptive spatial spectral processing may recover target energy spread over multiple pixels and reduce background clutter to increase the signal-to-noise ratio.

Growing community of inventors

Eatons Neck, NY, United States of America

Anthony M Sommese