Frank J. Crosby

Signature adaptive mine detection at a constant false alarm rate

A constant false alarm rate algorithm has been developed for use in multi-band mine detection. While it is often difficult to predict the spectral signatures of targets, the shape of the target may be known. This test exploits geometric target features and spectral differences between the target and the surrounding area. The algorithm is derived from a general statistical model of the data, which allows it to adapt to changing backgrounds and variable signatures.

Signature adaptive target detection and threshold selection for constant false alarm rate

A constant false alarm rate (CFAR) detection algorithm and a threshold selection algorithm are adapted and developed for use in multiband-image small-target detection. While it is often difficult to predict the spectral signatures of targets, the shape of the target may be known. This detection algorithm exploits geometric target features and spectral differences between the target and the surrounding area. The detection algorithm is derived from a general statistical model of the data with most emphasis on the background. The utility of CFAR algorithms is that the selection of a detection threshold can be made independently of image intensity. However, varied applications of the algorithms show that detection values are dependent on the scene adherence to the model. Achieving a CFAR in applications is very difficult. The threshold for a desired number of false alarms fluctuates with differing backgrounds. By appropriately mapping observations to the model, an automatic threshold selection algorithm is shown. Combining the CFAR-detection algorithm with the threshold selection algorithm produces a reliable constant false alarm rate.

Glint-induced false alarm reduction in signature adaptive target detection

The signal adaptive target detection algorithm developed by Crosby and Riley uses target geometry to discern anomalies in local backgrounds. Detection is not restricted based on specific target signatures. The robustness of the algorithm is limited by an increased false alarm potential. The base algorithm is extended to eliminate one common source of false alarms in a littoral environment. This common source is glint reflected on the surface of water. The spectral and spatial transience of glint prevent straightforward characterization and complicate exclusion. However, the statistical basis of the detection algorithm and its inherent computations allow for glint discernment and the removal of its influence.

Curvature Scale Space Application to Distorted Object Recognition and Classification

Contour classification methods which operate directly on an image are greatly affected by small magnitude transformations to the image. In this paper, a contour classification method is developed which takes advantage of curvature scale space (CS2) and a linear support vector machine (SVM) classifier. The CS2 representation boasts invariance to transformations including: scaling, rotation, translation and noise. In addition, the linear SVM is a robust tool for classification problems involving multiple labels. The combination of these tools produces a classifier well suited for object recognition in photographs where distortion is present.

Geometric correction through complex interpolation

A method for the correction of images that are subject to spatial warping is presented. The correction method applies to two-dimensional images and employs the construction of a complex-valued interpolating polynomial. The construction is based on defining a set of points in an image whose desired location is known. The coefficients of the polynomial are shown to satisfy a recursion relation that simplifies implementation and improves evaluation efficiency. Results are presented for a two-dimensional slice from a magnetic resonance image.

Total Variation Methods for Three Dimensional Lidar Image Denoising

New imaging capabilities have given rise to higher dimensional image processing. This paper presents a generalization of total variation (TV) based denoising model with a specific application to three-dimensional flash lidar imagery. The generalization uses a weighted norm, rather than the standard Euclidean measure, that accounts for sampling differences that may exist along different axes. We compare this new method against successive two-dimensional denoising and three-dimensional TV denoising. The results show that denoising with a weighted norm presents much better object recognition for multi-plane objects compared with both of the other methods. Further, it does not blur single plane objects beyond the result obtained with two-dimensional denoising.

Sequential approach to three-dimensional geometric image correction

This paper presents a new and comprehensive approach for correcting magnetic-resonance images that are subject to three-dimensional geometric distortion. Distortion in such images is typically caused by variations in the magnetic- field gradient in each of the three spatial dimensions. The new approach sequentially applies one-dimensional and two- dimensional correction techniques to achieve a complete three-dimensional geometric correction. It thus avoids many theoretical complications and computational inefficiencies that are inherently associated with direct (non-sequential) three-dimensional correction techniques.

Background adaptive multispectral band selection

AN initial automated band selection algorithm suitable for real-time application with tunable multispectral cameras is presented for multispectral target detection. The method and algorithm were developed from analyses of several background and target signatures collected from a field test using the prototype Tunable Filter Multispectral Camera (TFMC). Target and background data from TFMC imagery were analyzed to determine the detection performance of 32,768 unique 3-band channel combinations in the visible through and near IR spectral regions. This tuning knowledge base was analyzed to develop rules for an initial dynamic tuning algorithm. The performance data was sorted by conventional means to determine the best 3-band combinations. Methods were then developed to determine performance enhancing band sets for particular backgrounds and a variety of targets. This knowledge is then used in an algorithm to affect a real-time 3-band tuning capability. Additional band sets for real-time background categorization are chosen by both the ability to spectrally detect of one background from another. This work will illustrate an example of the performance results form the analysis for three targets on various backgrounds.

Compression for small targets in multispectral imagery

This paper analyzes the performance of a fast, low complexity, integer-to-integer compression scheme that is designed to give greater importance to small targets. Practical real-time operation of unmanned aerial vehicle mine/minefield detection systems has two difficult constraints. One is limited data-link bandwidth and the other is limited on-board processing power. Standard compression techniques are usually complex and tend to remove small objects from the imagery. In the imagery used for airborne mine/minefield detection, the targets are small, usually on the order of a few pixels. The region-of-interest (ROI) Wavelet Difference Reduction (WDR) compression scheme satisfies both of these con-straints and is shown to preserve detection rates of small targets. Results are compared for block-based (BB)-WDR compressed and ROI-WDR compressed and uncompressed images. The ROI -WDR process is shown to be superior to other compression conditions.

Airborne laser-diode-array illuminator assessment for the night vision's airborne mine-detection arid test

The Airborne Littoral Reconnaissance Technologies (ALRT) project has developed and tested a nighttime operational minefield detection capability using commercial off-the-shelf high-power Laser Diode Arrays (LDAs). The Coastal System Station’s ALRT project, under funding from the Office of Naval Research (ONR), has been designing, developing, integrating, and testing commercial arrays using a Cessna airborne platform over the last several years. This has led to the development of the Airborne Laser Diode Array Illuminator wide field-of-view (ALDAI-W) imaging test bed system. The ALRT project tested ALDAI-W at the Army’s Night Vision Lab’s Airborne Mine Detection Arid Test. By participating in Night Vision’s test, ALRT was able to collect initial prototype nighttime operational data using ALDAI-W, showing impressive results and pioneering the way for final test bed demonstration conducted in September 2003. This paper describes the ALDAI-W Arid Test and results, along with processing steps used to generate imagery.