David P. Williams

Exploiting Environmental Information for Improved Underwater Target Classification in Sonar Imagery

In many remote-sensing applications, measured data are a strong function of the environment in which they are collected. This paper introduces a new context-dependent classification algorithm to address and exploit this phenomenon. Within the proposed framework, an ensemble of classifiers is constructed, each associated with a particular environment. The key to the method is that the relative importance of each object (i.e., data point) during the learning phase for a given classifier is controlled via a modulating factor based on the similarity of auxiliary environment features. Importantly, the number of classifiers to learn and all other associated model parameters are inferred automatically from the training data. The promise of the proposed method is demonstrated on classification tasks seeking to distinguish underwater targets from clutter in synthetic aperture sonar imagery. The measured data were collected with an autonomous underwater vehicle during several large experiments, conducted at sea between 2008 and 2012, in different geographical locations with diverse environmental conditions. For these data, the environment was quantified by features (extracted from the imagery directly) measuring the anisotropy and the complexity of the seabed. Experimental results suggest that the classification performance of the proposed approach compares favorably to conventional classification algorithms as well as state-of-the-art context-dependent methods. Results also reveal the object features that are salient for performing target classification in different underwater environments.

Fast Target Detection in Synthetic Aperture Sonar Imagery: A New Algorithm and Large-Scale Performance Analysis

In this paper, a new unsupervised algorithm for the detection of underwater targets in synthetic aperture sonar (SAS) imagery is proposed. The method capitalizes on the high-quality SAS imagery whose high resolution permits many pixels on target. One particularly novel component of the method also detects sand ripples and estimates their orientation. The overall algorithm is made fast by employing a cascaded architecture and by exploiting integral-image representations. As a result, the approach makes near-real-time detection of proud targets in sonar data onboard an autonomous underwater vehicle (AUV) feasible. No training data are required because the proposed method is adaptively tailored to the environmental characteristics of the sensed data that are collected in situ. To validate and assess the performance of the proposed detection algorithm, a large-scale study of SAS images containing various mine-like targets is undertaken. The data were collected with the MUSCLE AUV during six large sea experiments, conducted between 2008 and 2012, in different geographical locations with diverse environmental conditions. The analysis examines detection performance as a function of target type, aspect, range, image quality, seabed environment, and geographical site. To our knowledge, this study-based on nearly 30 000 SAS images collectively covering approximately 160 km 2 of seabed, and involving over 1100 target detection opportunities-represents the most extensive such systematic, quantitative assessment of target detection performance with SAS data to date. The analysis reveals the variables that have the largest impact on target detection performance, namely, image quality and environmental conditions on the seafloor. Ways to exploit the results for adaptive AUV surveys using through-the-sensor data are also suggested.

Fast Unsupervised Seafloor Characterization in Sonar Imagery Using Lacunarity

A new unsupervised approach for characterizing seafloor in side-looking sonar imagery is proposed. The approach is based on lacunarity, which measures the pixel-intensity variation, of through-the-sensor data. No training data are required, no assumptions regarding the statistical distributions of the pixels are made, and the universe of (discrete) seafloor types need not be enumerated or known. It is shown how lacunarity can be computed very quickly using integral-image representations, thereby making real-time seafloor assessments on-board an autonomous underwater vehicle feasible. The promise of the approach is demonstrated on high-resolution synthetic-aperture-sonar imagery of diverse seafloor conditions measured at various geographical sites. Specifically, it is shown how lacunarity can effectively distinguish different seafloor conditions and how this fact can be exploited for target-detection performance prediction in mine-countermeasure operations.

Classification with imperfect labels for fault prediction

Classification techniques have been widely used in fault prediction for industrial systems. However, an inherent issue with this approach is label imperfections in training data, since the line of demarcation between classes is determined based on field expert experience and maintenance capability. To address this issue we propose a noisy-label model in which the labeling noise function is derived from a point of view motivated by reliability analysis. We also present a novel label bootstrapping method that can better reflect the true uncertainty of the labeling process than the standard approach for addressing label imperfections. The proposed technique gives encouraging results on two industrial fault-prediction data sets.

Adaptive underwater sonar surveys in the presence of strong currents

We consider the task of conducting underwater surveys with a sonar-equipped autonomous underwater vehicle (AUV) in environments with strong currents. More specifically, this topic is addressed in the context of mine countermeasure operations employing synthetic aperture sonar (SAS) sensors. Two complementary algorithms that allow the AUV to autonomously adapt its survey route based on sophisticated sensor data it collects in situ, while respecting the unique constraints imposed by the problem, are proposed. The algorithms allow the AUV to (i) adapt its survey heading based on the presence of currents to ensure quality data is collected, and (ii) adapt its survey route to reinspect the most suspicious objects at additional aspects. The flexibility to immediately react in situ to the environmental and tactical conditions sensed during the mission allow the most useful data for object recognition purposes to be collected efficiently. By obviating the recovery and redeployment of the AUV, as well as laboratory-based data-processing during the interregnum, the overall mission timeline can be greatly compressed and operational costs can be reduced. Experimental results illustrating the real-time execution of the proposed algorithms on an AUV are shown for a completely autonomous mission conducted in the North Sea.

Underwater target classification in synthetic aperture sonar imagery using deep convolutional neural networks

Deep convolutional neural networks are used to perform underwater target classification in synthetic aperture sonar (SAS) imagery. The deep networks are learned using a massive database of real, measured sonar data collected at sea during different expeditions in various geographical locations. A novel training procedure is developed specially for the data from this new sensor modality in order to augment the amount of training data available for learning and to avoid overfitting. The deep networks learned are employed for several binary classification tasks in which different classes of objects in real sonar data are to be discriminated. The proposed deep approach consistently achieves superior performance to a traditional feature-based classifier that we had relied on previously.

On Human Perception and Automatic Target Recognition: Strategies for Human-Computer Cooperation

This work addresses the task of underwater object recognition in sonar imagery when both human operators and automated algorithms are available. We discuss the issues that have impeded previous attempts at automation, raise key insights related to human perception, present strategies to exploit the skills of humans and computers synergistically, and demonstrate the utility of the proposed approaches on a real object-recognition task employing actual humans acting as operators. Importantly, the strategies outlined here can be immediately adopted in existing (unautomated) target recognition systems with minimal cost, effort, and risk, while still achieving potentially significant performance gains. Moreover, this progress lays the foundation for the acceptance of still-further automated systems in the future. Experimental results are provided from a real mine-search exercise at sea, with recognition performance as a function of human operator effort given for various human-computer divisions of labor.

Efficient dense sonar surveys with an autonomous underwater vehicle

An algorithm for the in situ adaptation of the survey route of an autonomous underwater vehicle (AUV) equipped with side-looking sonars was recently proposed. This algorithm immediately exploits the through-the-sensor data that is collected during a mission in order to ensure that quality data is collected everywhere in the area of interest. By introducing flexibility into the survey of the AUV, various limitations of pre-planned surveys are overcome. In particular, the need to re-deploy the AUV (to fill gaps in the data coverage) is obviated. In turn, the time and costs of the data-collection mission are significantly reduced. In this work, we improve the aforementioned algorithm by introducing an additional constraint to the survey track-selection process. This modification significantly increases the efficiency of a survey by further reducing both transit time and the overall number of tracks executed. In particular, the revised algorithm more closely approximates the optimal survey route that would be...

Multi-view SAS image classification using deep learning

A new approach is proposed for multi-view classification when sonar data is in the form of imagery and each object has been viewed an arbitrary number of times. An image-fusion technique is employed in conjunction with a deep learning algorithm (based on Boltzmann machines) so that the sonar data from multiple views can be combined and exploited at the (earliest) image level. The method utilizes single-view imagery and, whenever available, multi-view fused imagery, in the same unified classification framework. The promise of the proposed approach is demonstrated in the context of an object classification task with real synthetic aperture sonar (SAS) imagery collected at sea.

Multi-look processing of high-resolution SAS data for improved target detection performance

The rich content of synthetic aperture sonar (SAS) data is typically used to generate imagery with resolution as high as theoretically possible. But when the ultimate purpose of the imagery is for detecting objects with sizes large compared to the resolution cell, exploiting the raw data in alternative ways can be more useful. We first show how multiple lower-resolution SAS images (i.e., “looks”) can be obtained in an efficient, principled manner by band-limiting the image wavenumber spectrum of a full-resolution SAS image. By combining these multiple looks (from different aspect and frequency bands), a despeckled image that enjoys greater (target) signal to (seabed) reverberation ratio can be produced. On a large set of real SAS data collected at sea, it is demonstrated how better target detection performance of underwater mines can be achieved by using the lower-resolution despeckled imagery rather than single-look, maximum-resolution imagery.