Mine detection and classification using high-resolution sidescan sonar is a critical technology for mine counter measures (MCM). As opposed to the majority of techniques which require large training data sets, this paper presents unsupervised models for both the detection and the shadow extraction phases of an automated classification system. The detection phase is carried out using an unsupervised Markov random field (MRF) model where the required model parameters are estimated from the original image. Using a priori spatial information on the physical size and geometric signature of mines in sidescan sonar, a detection-orientated MRF model is developed which directly segments the image into regions of shadow, seabottom-reverberation, and object-highlight. After detection, features are extracted so that the object can be classified. A novel co-operating statistical snake (CSS) model is presented which extracts the highlight and shadow of the object. The CSS model again utilizes available a priori information on the spatial relationship between the highlight and shadow, allowing accurate segmentation of the object's shadow to be achieved.

/pdf/an-automatic-approach-to-the-detection-and-extraction-of-2fqnqbhw2h.pdf

An automatic approach to the detection and extraction of mine features in sidescan sonar

An electronically steered weather radar system comprises a plurality of transmit/receive modules, a plurality of antenna modules, and a system signal processor. Each transmit/receive module may be configured to adjust a phase characteristic of a radio frequency (RF) signal to be transmitted and received. Antenna modules may be in communication with transmit/receive modules and may form an antenna array configured to transmit a system beam in a direction determined by the phase of the RF signal from each transmit/receive module and to generate the RF signal from the received system beam. The system signal processor may be configured to perform a first scan to detect meteorological formations wherein the system signal processor communicates with each transmit/receive module to adjust a phase characteristic of the RF signals such that the antenna array transmits and receives the system beam through a plurality of azimuth angles and a plurality of elevation angles.

Electronically steered weather radar

The Poisson-binomial probability density function (pdf) describes the numbers of successes in N independent trials, when the individual probabilities of success vary across trials. Its use is pervasive in applications, such as fault tolerance, signal detection, target tracking, object classification/identification, multi-sensor data fusion, system management, and performance characterization, among others. We present a closed-form expression for this pdf, and we discuss several of its advantages regarding computing speed and implementation and in simplifying analysis, with examples of the latter including the computation of moments and the development of new trigonometric identities for the binomial coefficient and the binomial cumulative distribution function (cdf). Finally we also pose and address the inverse Poisson-binomial problem; that is, given such pdf, how to find (within a permutation) the probabilities of success of the individual trials.

Closed-Form Expression for the Poisson-Binomial Probability Density Function

An advanced capability for automated detection and classification of sea mines in sonar imagery has been developed. The advanced mine detection and classification (AMDAC) algorithm consists of an improved detection density algorithm, a classification feature extractor that uses a stepwise feature selection strategy, a k-nearest neighbor attractor-based neural network (KNN) classifier, and an optimal discriminatory filter classifier. The detection stage uses a nonlinear matched filter to identify mine-size regions in the sonar image that closely match a mine's signature. For each detected mine-like region, the feature extractor calculates a large set of candidate classification features. A stepwise feature selection process then determines the subset features that optimizes probability of detection and probability of classification for each of the classifiers while minimizing false alarms.

Automated detection and classification of sea mines in sonar imagery

This compilation was prepared for the Dudley Knox Library, Naval Postgraduate School, Monterey, CA

https://calhoun.nps.edu/bitstream/10945/6976/1/Jul09-Seamines_biblio.pdf

Sea Mines and Countermeasures: A Bibliography

An automatic, robust, adaptive clutter suppression, mine detection and classification processing string has been developed and applied to side-scan sonar imagery data. The overall processing string includes data pre-processing, adaptive clutter filtering (ACF), 2D normalization, detection, feature extraction, and classification processing blocks. The data pre-processing block contains automatic gain control and data decimation processing. The ACF technique designs a 2D adaptive range-crossrange linear FIR filter which is optimal in the Least Squares sense, simultaneously suppressing the background clutter while preserving an average peak target signature (normalized shape) computed a priori using training set data. A multiple reference ACF algorithm version was utilized to account for multiple target shapes (due to different mine types, multiple target aspect angles, etc.). The detection block consists of thresholding, clustering of exceedances and limiting their number, and a secondary thresholding process. Following feature extraction, the classification block applies a novel transformation to the data, which orthogonalizes the features and enables an efficient application of the optimal log-likelihood-ratio-test (LLRT) classification rule. The utility of the overall processing string was demonstrated with two side-scan sonar data sets. The ACF/feature orthogonalization based LLRT mine classification processing string provided average probability of correct mine classification and false alarm rate performance similar to that obtained when utilizing an expert sonar operator.

Adaptive-filter/feature-orthogonalization processing string for optimal LLRT mine classfication in side-scan sonar imagery

An automatic, robust, adaptive clutter suppression, predetection level fusion, sea mine detection and classification processing string has been developed and applied to shallow water side-scan sonar imagery data. The overall processing string includes pre-processing string includes pre-processing, adaptive clutter filtering (ACF), 2D normalization, detection, feature extraction and classification processing blocks. The pre-processing block contains automatic gain control, data decimation and data alignment processing. The ACF is a multi-dimensional adaptive linear FIR filter, optimal in the least squares sense, for simultaneous background clutter suppression and preservation of an average peak target signature. After data alignment, using a 3D ACF enables simultaneous multiple frequency data fusion and clutter suppression in the composite frequency-range-crossrange domain. Following 2D normalization, the detection consists of thresholding, clustering of exceedances and limiting their number. Finally, features are extracted and a orthogonalization transformation is applied to the data, enabling an efficient application of the optimal log-likelihood-ratio-test (LLRT) classification rule. The utility of the overall processing string was demonstrated with two side-scan sonar data sets. The ACF, feature orthogonalization, LLRT-based classification processing string provided average probability of correct mine classification and false alarm rate performance exceeding the one obtained when utilizing an expert sonar operator. The overall processing string can be easily implemented in real-time using COTS technology.

Adaptive three-dimensional range-crossrange-frequency filter processing string for sea mine classification in side scan sonar imagery

An advanced, automatic, adaptive clutter suppression, sea mine detection, classification and fusion processing string has been developed and tested with high resolution sonar imagery dat. The overall computer-aided-detection/computer- aided-classification (CAD/CAC) string includes pre- processing, adaptive clutter filtering (ACF), normalization, detection, feature extraction, feature orthogonalization, subset feature selection, classification and fusion processing blocks. The ACF is an adaptive linear FIR filter, optimal in the Least Squares (LS) sense, and is applied to low-resolution data. Data pre-normalization, clipping and mean subtraction, allows application of a range dimension only ACF that is matched both to average highlight and shadow information, while simultaneously suppressing background clutter. Following post-ACF normalization, and detection consists of thresholding, clustering of exceedances and limiting the number of detections. Subsequently, features are extracted from high-resolution data and an orthogonalization transformation is applied to the features, enabling an efficient application of the optimal log-likelihood-ratio-test (LLRT) classification rule. Finally, the classified objects of three processing strings, developed by three different researchers, are fused, using an LLRT-based fusion rule. Processing string improvements have been developed over previous CAD/CAC and fusion string versions. The utility of the overall processing strings and their fusion was demonstrated with very shallow water high-resolution sonar imagery data sets, form a difficult environment. The processing string classification performance was optimized by appropriately selecting a subset of the original feature set. The fusion of the CAD/CAC processing strings resulted in improved mine classification capability, providing a three-fold false alarm rate reduction, compared to the best individual CAD/CAC processing string results.

Side-scan sonar imagery fusion for sea mine detection and classification in very shallow water

A new sea mine computer-aided-detection/computer-aided-classification (CAD/CAC) processing string has been developed. The CAD/CAC processing string consists of preprocessing, adaptive clutter filtering (ACF), normalization, detection, feature extraction, classification and fusion processing blocks. The range-dimension ACF is an adaptive linear FIR filter, which is matched both to average highlight and shadow information, while also simultaneously suppressing background clutter. For each detected object, features are extracted and processed through an orthogonalization transformation, enabling an efficient application of the optimal log-likelihood-ratio-test (LLRT) classification rule, in the orthogonal feature space domain. The classified objects of 3 distinct processing strings, developed by 3 different researchers, are fused, using the classification confidence values as features and logic-based, M-out-of-N, or novel LLRT-based fusion rules. The utility of the overall processing strings and their fusion was demonstrated with new very shallow water high-resolution sonar imagery data. The processing string detection and classification parameters were tuned and the string classification performance was optimized, by appropriately selecting a subset of the original feature set. It was shown that LLRT-based fusion algorithms outperform the logic-based or the M-out-of-N ones. The LLRT-based fusion of the CAD/CAC processing strings resulted in a four-fold false alarm rate reduction, compared to the best single CAD/CAC processing string results, while maintaining a constant correct mine classification probability.

Fusion of adaptive algorithms for the classification of sea mines using high resolution side scan sonar in very shallow water

A need exists to develop robust automatic techniques for discriminating between minelike target and clutter returns in sonar imagery. To address this need, an adaptive clutter suppression linear FIR filtering technique has been developed and applied to side scan sonar imagery data. The adaptive filtering procedure consists of four stages. First, a normalized average target signature (shape) within the filter window is computed using training set data. Second, the background clutter covariance matrix is computed by scanning the filter window over the data. Third, following substitutions of the average target signature and covariance expressions into a set of normal equations, an adaptive filter is computed which simultaneously suppresses the background clutter while preserving the peak of the average target signature. Finally, the data is filtered using the 2D adaptive range-crossrange filter. The overall mine detection processing string includes automatic gain control, data decimation, adaptive clutter filtering (ACF), 2D normalization, thresholding, exceedance clustering, limiting the number of exceedances and secondary thresholding processing blocks. The utility of the ACF processing string was demonstrated with three side scan sonar datasets. The ACF algorithm provided average probability of detection and false alarm rate performance similar to that obtained when utilizing an expert sonar operator.

Manuel F. Fernandez

Papers

Adaptive-filter/feature-orthogonalization processing string for optimal LLRT mine classfication in side-scan sonar imagery

Adaptive three-dimensional range-crossrange-frequency filter processing string for sea mine classification in side scan sonar imagery

Side-scan sonar imagery fusion for sea mine detection and classification in very shallow water

Fusion of adaptive algorithms for the classification of sea mines using high resolution side scan sonar in very shallow water

Adaptive filter for mine detection and classification in side-scan sonar imagery