A framework for the fusion of computer-aided detection and classification algorithms for side-scan imagery is presented. The framework is based on the Dempster-Shafer theory of evidence, which permits fusion of heterogeneous outputs of target detectors and classifiers. The utilisation of augmented reality for the training and evaluation of the algorithms used over a large test set permits the optimisation of their performance. In addition, this framework is adaptive regarding two aspects. First, it allows for the addition of contextual information to the decision process, giving more importance to the outputs of those algorithms that perform better in particular mission conditions. Secondly, the fusion parameters are optimised on-line to correct for mistakes, which occur while deployed.

A proof of concept for a model-less target detection and classification system for side-scan imagery is presented. The system is based on a supervised approach that uses augmented reality (AR) images for training computer added detection and classification (CAD/CAC) algorithms, which are then deployed on real data. The algorithms are able to generalise and detect real targets when trained on AR ones, with performances comparable with the state-of-the-art in CAD/CAC. To illustrate the approach, the focus is on one specific algorithm, which uses Bayesian decision and the novel, purpose-designed central filter feature extractors. Depending on how the training database is partitioned, the algorithm can be used either for detection or classification. Performance figures for these two modes of operation are presented, both for synthetic and real targets. Typical results show a detection rate of more that 95% and a false alarm rate of less than 5%. The proposed supervised approach can be directly applied to train and evaluate other learning algorithms and data representations. In fact, a most important aspect is that it enables the use of a wealth of legacy pattern recognition algorithms for the sonar CAD/CAC applications of target detection and target classification.

This paper presents a framework for the fusion of detection and classification algorithms for side-scan imagery. The framework is based on Dempster-Shafer theory of evidence, which permits the fusion of heterogeneous outputs of targets detectors and classifiers. The paper will illustrate how the technique permits the incorporation of contextual information into the decision process, giving more importance to the outputs of those algorithms that perform better in particular mission conditions.

We describe the physical-optics modelling of a millimetre-wave imaging system intended to enable automated detection of threats hidden under clothes. This paper outlines the theoretical basis of the formation of millimetre-wave images and provides the model of the simulated imaging system. Results of simulated images are presented and the validation with real ones is carried out. Finally, we present a brief study of the potential materials to be classified in this system.

The accurate detection and identification of underwater targets continues as a major issue, despite, or perhaps as a result of, the promise of higher resolution underwater imaging systems, including synthetic aperture sonar and high frequency sidescan. Numerous techniques have been proposed for computer aided detection to detect all possible mine-like objects, and computer aided classification to classify whether the detected object is a target or not. The majority of existing techniques employ supervised classification systems which are reliant on training data. The success of these systems can be highly dependant on the similarity of the test data to the training data, which includes the effect of the background region on which the target was located. This paper will briefly discuss and compare two possible solutions to this problem. The first is a model based system for classification and the second utilises an augmented reality simulator to produce training data.

It is possible to reduce the error rate of a single classifier using a classifier ensemble. However, any gain in performance is undermined by the increased computation of performing classification several times. Here the AdaboostFS algorithm is proposed which builds on two popular areas of ensemble research: Adaboost and Ensemble Feature Selection (EFS). The aim of AdaboostFS is to reduce the number of features used by each base classifer and hence the overall computation required by the ensemble. To do this the algorithm combines a regularised version of Boosting AdaboostReg [1] with a floating feature search for each base classifier. AdaboostFS is compared using four benchmark data sets to AdaboostAll, which uses all features and to AdaboostRSM, which uses a random selection of features. Performance is assessed based on error rate, ensemble error and diversity, and the total number of features used for classification. Results show that AdaboostFS achieves a lower error rate and higher diversity than AdaboostAll, and achieves a lower error rate and comparable diversity to AdaboostRSM. However, over the other methods AdaboostFS produces a significant reduction in the number of features required for classification in each base classifier and the entire ensemble.

The ATRIUM project aims to the automatic detection of threats hidden under clothes using millimetre-wave imaging. We describe a simulator of realistic millimetre-wave images and a system for detecting metallic weapons automatically. The latter employs two stages, detection and tracking. We present a detector for metallic objects based on mixture models, and a target tracker based on particle filtering. We show convincing, simulated millimetre-wave images of the human body with and without hidden threats, including a comparison with real images, and very good detection and tracking performance with eight real sequences. (International Workshop on Pattern Recognition for Crime Prevention, Security and Surveillance)

Accurate measurements of the locations of surfacing cetaceans (whales, dolphins and porpoises) are important data for behavioral studies and sightings surveys. A system for tracking cetacean movements based on photogrammetric analysis of digital images, presented in the paper by R. Leaper and J. Gordon (2001), has been developed and tested at sea. This paper presents and discusses the use of image processing tools to partially automate the processing of the digital images thus produced, in order to obtain estimates of both the bearing and range of the sightings. It is hoped that these tools will be enablers of wider scale surveys, and they are expected to be deployed and field tested during a trial in the spring/summer 2005.

The use of video as an underwater sensor is wide spread in underwater communities. However, the automated processing and analysis of video data is only emerging. The underwater community can draw from an important legacy of video/optical image processing algorithms developed for other contexts, but little methodology on how to assess their suitability for underwater images currently exists. This paper suggests a methodology to quantitatively assess the robustness and behavior of algorithms in the face of underwater noises. It relies on the systematic simulation of underwater specific perturbations of images, with varying degrees of severity. The methodology enables to benchmark algorithms' suitability for underwater conditions.

This paper presents an experimental protocol developed for the design, performance estimation and comparison of underwater video classifier systems. Such systems have to be designed using application data that is small, sparse and extremely variable. The proposed protocol uses outlier rejection, data pairing, Bootstrap performance estimation and hypothesis testing to achieve a robust performance estimate and comparison between classifier designs. The protocol is demonstrated and assessed on an application experiment. The application involves the design of a classification system for the automated detection of trawling marks from mission video. Two systems are proposed using selective and geometric feature types and an ensemble classifier. The protocol robustly identifies differences between the two proposed system designs using error and discrimination rates. Overall the geometric feature system is chosen as the final system. The protocol was also compared with other performance estimates and found to have the closest match to actual test data performance.

This paper presents a study of the Boosting Feature Selection (BFS) algorithm [1], a method which incorporates feature selection into Adaboost. Such an algorithm is interesting as it combines the methods studied by Boosting and ensemble feature selection researchers. Observations are made on generalisation, weighted error and error diversity to compare the algorithms performance to Adaboost while using a nearest mean base learner. Ensemble feature prominence is proposed as a stop criterion for ensemble construction. Its quality assessed using the former performance measures. BFS is found to compete with Adaboost in terms of performance, despite the reduced feature description for each base classifer. This is explained using weighted error and error diversity. Results show the proposed stop criterion to be useful for trading ensemble performance and complexity.

Video-frame-rate millimetre-wave imaging has recently been demonstrated with a quality similar to that of a low-quality uncooled thermal imager. In this paper we will discuss initial investigations into the transfer of image processing algorithms from more mature imaging modalities to millimetre-wave imagery. The current aim is to develop body segmentation algorithms for use in object detection and analysis. However, this requires a variety of image processing algorithms from different domains, including image de-noising, segmentation and motion tracking. This paper focuses on results from the segmentation of a body from the millimetre-wave images and a qualitative comparison of different approaches is presented. Their performance is analysed and any characteristics which enhance or limit their application are discussed. While it is possible to apply image processing algorithms developed for the visible-band directly to millimetre-wave images, the physics of the image formation process is very different. This paper discusses the potential for exploiting an understanding of the physics of image formation in the image segmentation process to enhance classification of scene components and, thereby, improve segmentation performance. This paper presents some results from a millimetre-wave image formation simulator, including synthetic images with multiple objects in the scene.

Support vector machines (SVMs) are both mathematically well-funded and efficient in a large number of real-world applications. However, the classification results highly depend on the parameters of the model: the scale of the kernel and the regularization parameter. Estimating these parameters is referred to as tuning. Tuning requires to estimate the generalization error and to find its minimum over the parameter space. Classical methods use a local minimization approach. After empirically showing that the tuning of parameters presents local minima, we investigate in this paper the use of global minimization techniques, namely genetic algorithms and simulated annealing. This latter approach is compared to the standard tuning frameworks and provides a more reliable tuning method.

Motion estimation is a key problem in the analysis of image sequences. From a sequence of images we can only estimate an approximation of the image motion field called optical flow. We propose to improve optical flow estimation by including information from images of textural features. We compute the optical flow from intensity and textural images from first-order derivatives, then combine estimates using the spatial gradient as confidence measure. Experimental results with images for which the ground-truth optical flow is known show clearly that the estimate improves by including estimates from textural images. Experiments with several underwater images also show a qualitative improvement.

It is often the case that only a few sparse sequences of long videos from scientific underwater surveys actually contain important information for the expert. Locating such sequences is time consuming and tedious. A system that automatically detects those critical parts, online or during post-mission tape analysis, would alleviate the expert workload and improve data exploitation. In this paper, a methodology for evaluating the performance of such a system on real data is presented. Interesting sequences are started by changes of visual context. An algorithm to detect significant context changes in benthic videos in real time has been presented by Lebart et al. in 2000. It is used as an illustration for this methodology-its performance is studied and benchmarked on real underwater data, ground truthed by an expert biologist. Various issues relating to the complexity of the problems of automatically analyzing underwater video are also discussed.