One of the desired capabilities for wide-area persistent ISR systems is to reliably locate and subsequently track the movement of targets within the field of view. Current wide-area persistent ISR systems are characterized by large pixel overall counts and very large fields of view. This leads to a large ground sample distance with few pixels-on-target. Locating targets under these constraints is extremely difficult due to the fact that the targets present very little detailed structure. In this paper we will present the application of rich image feature descriptors combined with advanced statistical target detection methodologies to the airborne ISR problem. We will demonstrate that these algorithms can reliably locate targets in the scene without relying on the target's motion to form a detection. This is useful in ISR application where it is desirable to be able to continuously track a target through stops and maneuvers.
Short wave infrared (SWIR) spectral imaging systems are vital for Intelligence, Surveillance, and Reconnaissance (ISR)
applications because of their abilities to autonomously detect targets and classify materials. Typically the spectral
imagers are incapable of providing Full Motion Video (FMV) because of their reliance on line scanning. We enable
FMV capability for a SWIR multi-spectral camera by creating a repeating pattern of 3x3 spectral filters on a staring focal
plane array (FPA). In this paper we present the imagery from an FMV SWIR camera with nine discrete bands and
discuss image processing algorithms necessary for its operation. The main task of image processing in this case is
demosaicking of the spectral bands i.e. reconstructing full spectral images with original FPA resolution from spatially
subsampled and incomplete spectral data acquired with the choice of filter array pattern. To the best of author's
knowledge, the demosaicking algorithms for nine or more equally sampled bands have not been reported before.
Moreover all existing algorithms developed for demosaicking visible color filter arrays with less than nine colors assume
either certain relationship between the visible colors, which are not valid for SWIR imaging, or presence of one color
band with higher sampling rate compared to the rest of the bands, which does not conform to our spectral filter pattern.
We will discuss and present results for two novel approaches to demosaicking: interpolation using multi-band edge
information and application of multi-frame super-resolution to a single frame resolution enhancement of multi-spectral
spatially multiplexed images.
Traditional Fabry-Perot (FP) spectroscopy is bandwidth limited to avoid mixing signals from different transmission
orders of the interferometer. Unlike Fourier transformation, the extraction of spectra from multiple-order interferograms
resulting from multiplexed optical signals is in general an ill-posed problem. Using a Fourier transform approach, we
derive a generalized Nyquist limit appropriate to signal recovery from FP interferograms. This result is used to derive a
set of design rules giving the usable wavelength range and spectral resolution of FP interferometers or etalon arrays
given a set of accessible physical parameters. Numerical simulations verify the utility of these design rules for moderate
resolution spectroscopy with bandwidths limited by the detector spectral response. Stable and accurate spectral recovery
over more than one octave is accomplished by simple matrix multiplication of the interferogram. In analogy to recently
developed single-order micro-etalon arrays (Proc. of SPIE v.8266, no. 82660Q), we introduce Multiple-Order Staircase
Etalon Spectroscopy (MOSES), in which micro-arrays of multiple order etalons can be bonded to or co-fabricated with a
sensor array. MOSES enables broader bandwidth multispectral and hyperspectral instruments than single-order etalon
arrays while keeping a physical footprint insignificantly different from that of the detection array.
We report on the application of Optical Flow (OF) and state-of-the art multi-frame Super-Resolution (SR) algorithms to imagery that models space objects (SOs). Specifically, we demonstrate the ability to track SOs through sequences consisting of tens of images using different OF algorithms and show dependence of the tracking accuracy on illumination condition changes and on the values of pixel displacements between neighboring images. Additionally, we demonstrate spatial acuity enhancement of the pixel limited resolution of SO motion imagery by applying a novel SR algorithm accounting for OF errors.
The largest challenge with all persistent surveillance systems is they require a trade between area coverage and ground object resolution. This trade typically results in provision of imagery where objects desired to be tracked have a small total number of pixels (often less than a few hundred total). With such low pixel counts, traditional target recognition methods become difficult. For this reason, most persistent surveillance tracking systems are based on detection and tracking of image changes. These change-detection tracking systems, however, struggle to maintain tracks through quick maneuvers, stops, obscurations, and dense traffic. Feature descriptors, including template matching, histogram of oriented gradients (HOG), and local binary patterns (LBP) are evaluated for use in the special case of very low pixel count target detection and track maintenance. These dynamic feature-based detection models are incorporated into a change-detection based tracking system. The resulting composite tracking system will be described as applied to EO and MWIR wide area data collected under a variety of conditions. Resulting tracking system improvements and tradeoffs between feature descriptors are presented.