Unusual scene detection using distributed behaviour model and sparse representation

The ability to detect unusual events in surviellance footage as they happen is a highly desireable feature for a surveillance system. However, this problem remains challenging in crowded scenes due to occlusions and the clustering of people. In this paper, we propose using the Distributed Behavior Model (DBM), which has been widely used in computer graphics, for video event detection. Our approach does not rely on object tracking, and is robust to camera movements. We use sparse coding for classification, and test our approach on various datasets. Our proposed approach outperforms a state-of-the-art work which uses the social force model and Latent Dirichlet Allocation.

Fusion of morphological images for airborne target detection

Several track-before-detection approaches for image based aircraft detection have recently been examined in an important automated aircraft collision detection application. A particularly popular approach is a two stage processing paradigm which involves: a morphological spatial filter stage (which aims to emphasize the visual characteristics of targets) followed by a temporal or track filter stage (which aims to emphasize the temporal characteristics of targets). In this paper, we proposed new spot detection techniques for this two stage processing paradigm that fuse together raw and morphological images or fuse together various different morphological images (we call these approaches morphological reinforcement). On the basis of flight test data, the proposed morphological reinforcement operations are shown to offer superior signal to-noise characteristics when compared to standard spatial filter options (such as the close-minus-open and adaptive contour morphological operations). However, system operation characterised curves, which examine detection verses false alarm characteristics after both processing stages, illustrate that system performance is very data dependent.

HMM triangle relative entropy concepts in sequential change detection applied to vision-based dim target manoeuvre detection

The quick detection of abrupt (unknown) parameter changes in an observed hidden Markov model (HMM) is important in several applications. Motivated by the recent application of relative entropy concepts in the robust sequential change detection problem (and the related model selection problem), this paper proposes a sequential unknown change detection algorithm based on a relative entropy based HMM parameter estimator. Our proposed approach is able to overcome the lack of knowledge of post-change parameters, and is illustrated to have similar performance to the popular cumulative sum (CUSUM) algorithm (which requires knowledge of the post-change parameter values) when examined, on both simulated and real data, in a vision-based aircraft manoeuvre detection problem.

Spectrum sensing, detection and optimisation in cognitive radio for non-stationary primary user signals

Cognitive radio is an emerging technology proposing the concept of dynamic spec- trum access as a solution to the looming problem of spectrum scarcity caused by the growth in wireless communication systems. Under the proposed concept, non- licensed, secondary users (SU) can access spectrum owned by licensed, primary users (PU) so long as interference to PU are kept minimal. Spectrum sensing is a crucial task in cognitive radio whereby the SU senses the spectrum to detect the presence or absence of any PU signal. Conventional spectrum sensing assumes the PU signal as ‘stationary’ and remains in the same activity state during the sensing cycle, while an emerging trend models PU as ‘non-stationary’ and undergoes state changes. Existing studies have focused on non-stationary PU during the transmission period, however very little research considered the impact on spectrum sensing when the PU is non-stationary during the sensing period. The concept of PU duty cycle is developed as a tool to analyse the performance of spectrum sensing detectors when detecting non-stationary PU signals. New detectors are also proposed to optimise detection with respect to duty cycle ex- hibited by the PU. This research consists of two major investigations. The first stage investigates the impact of duty cycle on the performance of existing detec- tors and the extent of the problem in existing studies. The second stage develops new detection models and frameworks to ensure the integrity of spectrum sensing when detecting non-stationary PU signals. The first investigation demonstrates that conventional signal model formulated for stationary PU does not accurately reflect the behaviour of a non-stationary PU. Therefore the performance calculated and assumed to be achievable by the conventional detector does not reflect actual performance achieved. Through analysing the statistical properties of duty cycle, performance degradation is proved to be a problem that cannot be easily neglected in existing sensing studies when PU is modelled as non-stationary. The second investigation presents detectors that are aware of the duty cycle ex- hibited by a non-stationary PU. A two stage detection model is proposed to improve the detection performance and robustness to changes in duty cycle. This detector is most suitable for applications that require long sensing periods. A second detector, the duty cycle based energy detector is formulated by integrat- ing the distribution of duty cycle into the test statistic of the energy detector and suitable for short sensing periods. The decision threshold is optimised with respect to the traffic model of the PU, hence the proposed detector can calculate average detection performance that reflect realistic results. A detection framework for the application of spectrum sensing optimisation is proposed to provide clear guidance on the constraints on sensing and detection model. Following this framework will ensure the signal model accurately reflects practical behaviour while the detection model implemented is also suitable for the desired detection assumption. Based on this framework, a spectrum sensing optimisation algorithm is further developed to maximise the sensing efficiency for non-stationary PU. New optimisation constraints are derived to account for any PU state changes within the sensing cycle while implementing the proposed duty cycle based detector.

A practical signal processing approach for condition monitoring of low speed machinery using Peak-Hold-Down-Sample algorithm

A simple and effective down-sample algorithm, Peak-Hold-Down-Sample (PHDS) algorithm is developed in this paper to enable a rapid and efficient data transfer in remote condition monitoring applications. The algorithm is particularly useful for high frequency Condition Monitoring (CM) techniques, and for low speed machine applications since the combination of the high sampling frequency and low rotating speed will generally lead to large unwieldy data size. The effectiveness of the algorithm was evaluated and tested on four sets of data in the study. One set of the data was extracted from the condition monitoring signal of a practical industry application. Another set of data was acquired from a low speed machine test rig in the laboratory. The other two sets of data were computer simulated bearing defect signals having either a single or multiple bearing defects. The results disclose that the PHDS algorithm can substantially reduce the size of data while preserving the critical bearing defect information for all the data sets used in this work even when a large down-sample ratio was used (i.e., 500 times down-sampled). In contrast, the down-sample process using existing normal down-sample technique in signal processing eliminates the useful and critical information such as bearing defect frequencies in a signal when the same down-sample ratio was employed. Noise and artificial frequency components were also induced by the normal down-sample technique, thus limits its usefulness for machine condition monitoring applications.

Efficient real-time face detection for high resolution surveillance applications

This paper presents an efficient face detection method suitable for real-time surveillance applications. Improved efficiency is achieved by constraining the search window of an AdaBoost face detector to pre-selected regions. Firstly, the proposed method takes a sparse grid of sample pixels from the image to reduce whole image scan time. A fusion of foreground segmentation and skin colour segmentation is then used to select candidate face regions. Finally, a classifier-based face detector is applied only to selected regions to verify the presence of a face (the Viola-Jones detector is used in this paper). The proposed system is evaluated using 640 x 480 pixels test images and compared with other relevant methods. Experimental results show that the proposed method reduces the detection time to 42 ms, where the Viola-Jones detector alone requires 565 ms (on a desktop processor). This improvement makes the face detector suitable for real-time applications. Furthermore, the proposed method requires 50% of the computation time of the best competing method, while reducing the false positive rate by 3.2% and maintaining the same hit rate.

Improved foreground detection via block-based classifier cascade with probabilistic decision integration

Background subtraction is a fundamental low-level processing task in numerous computer vision applications. The vast majority of algorithms process images on a pixel-by-pixel basis, where an independent decision is made for each pixel. A general limitation of such processing is that rich contextual information is not taken into account. We propose a block-based method capable of dealing with noise, illumination variations, and dynamic backgrounds, while still obtaining smooth contours of foreground objects. Specifically, image sequences are analyzed on an overlapping block-by-block basis. A low-dimensional texture descriptor obtained from each block is passed through an adaptive classifier cascade, where each stage handles a distinct problem. A probabilistic foreground mask generation approach then exploits block overlaps to integrate interim block-level decisions into final pixel-level foreground segmentation. Unlike many pixel-based methods, ad-hoc postprocessing of foreground masks is not required. Experiments on the difficult Wallflower and I2R datasets show that the proposed approach obtains on average better results (both qualitatively and quantitatively) than several prominent methods. We furthermore propose the use of tracking performance as an unbiased approach for assessing the practical usefulness of foreground segmentation methods, and show that the proposed approach leads to considerable improvements in tracking accuracy on the CAVIAR dataset.

SERS substrate for detection of explosives

A novel gold coated femtosecond laser nanostructured sapphire surface – an “optical nose” - based on surface-enhanced Raman spectroscopy (SERS) for detecting vapours of explosive substances was investigated. Four different nitroaromatic vapours at room temperature were tested. Sensor responses were unambiguous and showed response in the range of 0.05 – 15 uM at 25 °C. The laser fabricated substrate nanostructures produced up to an eight-fold increase in Raman signal over that observed on the unstructured portions of the substrate. This work demonstrates a simple sensing system that is compatible with commercial manufacturing practices to detect taggants in explosives which can undertake as part of an integrated security or investigative mission.

Collaborative intrusion detection framework : characteristics, adversarial opportunities and countermeasures

Complex Internet attacks may come from multiple sources, and target multiple networks and technologies. Nevertheless, Collaborative Intrusion Detection Systems (CIDS) emerges as a promising solution by using information from multiple sources to gain a better understanding of objective and impact of complex Internet attacks. CIDS also help to cope with classical problems of Intrusion Detection Systems (IDS) such as zero-day attacks, high false alarm rates and architectural challenges, e. g., centralized designs exposing the Single-Point-of-Failure. Improved complexity on the other hand gives raise to new exploitation opportunities for adversaries. The contribution of this paper is twofold. We first investigate related research on CIDS to identify the common building blocks and to understand vulnerabilities of the Collaborative Intrusion Detection Framework (CIDF). Second, we focus on the problem of anonymity preservation in a decentralized intrusion detection related message exchange scheme. We use techniques from design theory to provide multi-path peer-to-peer communication scheme where the adversary can not perform better than guessing randomly the originator of an alert message.

Decentralized detector generation in cooperative intrusion detection system

We consider Cooperative Intrusion Detection System (CIDS) which is a distributed AIS-based (Artificial Immune System) IDS where nodes collaborate over a peer-to-peer overlay network. The AIS uses the negative selection algorithm for the selection of detectors (e.g., vectors of features such as CPU utilization, memory usage and network activity). For better detection performance, selection of all possible detectors for a node is desirable but it may not be feasible due to storage and computational overheads. Limiting the number of detectors on the other hand comes with the danger of missing attacks. We present a scheme for the controlled and decentralized division of detector sets where each IDS is assigned to a region of the feature space. We investigate the trade-off between scalability and robustness of detector sets. We address the problem of self-organization in CIDS so that each node generates a distinct set of the detectors to maximize the coverage of the feature space while pairs of nodes exchange their detector sets to provide a controlled level of redundancy. Our contribution is twofold. First, we use Symmetric Balanced Incomplete Block Design, Generalized Quadrangles and Ramanujan Expander Graph based deterministic techniques from combinatorial design theory and graph theory to decide how many and which detectors are exchanged between which pair of IDS nodes. Second, we use a classical epidemic model (SIR model) to show how properties from deterministic techniques can help us to reduce the attack spread rate.

Social signal processing for pain monitoring using a hidden conditional random field

utomatic pain monitoring has the potential to greatly improve patient diagnosis and outcomes by providing a continuous objective measure. One of the most promising methods is to do this via automatically detecting facial expressions. However, current approaches have failed due to their inability to: 1) integrate the rigid and non-rigid head motion into a single feature representation, and 2) incorporate the salient temporal patterns into the classification stage. In this paper, we tackle the first problem by developing a “histogram of facial action units” representation using Active Appearance Model (AAM) face features, and then utilize a Hidden Conditional Random Field (HCRF) to overcome the second issue. We show that both of these methods improve the performance on the task of pain detection in sequence level compared to current state-of-the-art-methods on the UNBC-McMaster Shoulder Pain Archive.

Speaker diarization : "who spoke when"

Speaker diarization is the process of annotating an input audio with information that attributes temporal regions of the audio signal to their respective sources, which may include both speech and non-speech events. For speech regions, the diarization system also specifies the locations of speaker boundaries and assign relative speaker labels to each homogeneous segment of speech. In short, speaker diarization systems effectively answer the question of ‘who spoke when’. There are several important applications for speaker diarization technology, such as facilitating speaker indexing systems to allow users to directly access the relevant segments of interest within a given audio, and assisting with other downstream processes such as summarizing and parsing. When combined with automatic speech recognition (ASR) systems, the metadata extracted from a speaker diarization system can provide complementary information for ASR transcripts including the location of speaker turns and relative speaker segment labels, making the transcripts more readable. Speaker diarization output can also be used to localize the instances of specific speakers to pool data for model adaptation, which in turn boosts transcription accuracies. Speaker diarization therefore plays an important role as a preliminary step in automatic transcription of audio data. The aim of this work is to improve the usefulness and practicality of speaker diarization technology, through the reduction of diarization error rates. In particular, this research is focused on the segmentation and clustering stages within a diarization system. Although particular emphasis is placed on the broadcast news audio domain and systems developed throughout this work are also trained and tested on broadcast news data, the techniques proposed in this dissertation are also applicable to other domains including telephone conversations and meetings audio. Three main research themes were pursued: heuristic rules for speaker segmentation, modelling uncertainty in speaker model estimates, and modelling uncertainty in eigenvoice speaker modelling. The use of heuristic approaches for the speaker segmentation task was first investigated, with emphasis placed on minimizing missed boundary detections. A set of heuristic rules was proposed, to govern the detection and heuristic selection of candidate speaker segment boundaries. A second pass, using the same heuristic algorithm with a smaller window, was also proposed with the aim of improving detection of boundaries around short speaker segments. Compared to single threshold based methods, the proposed heuristic approach was shown to provide improved segmentation performance, leading to a reduction in the overall diarization error rate. Methods to model the uncertainty in speaker model estimates were developed, to address the difficulties associated with making segmentation and clustering decisions with limited data in the speaker segments. The Bayes factor, derived specifically for multivariate Gaussian speaker modelling, was introduced to account for the uncertainty of the speaker model estimates. The use of the Bayes factor also enabled the incorporation of prior information regarding the audio to aid segmentation and clustering decisions. The idea of modelling uncertainty in speaker model estimates was also extended to the eigenvoice speaker modelling framework for the speaker clustering task. Building on the application of Bayesian approaches to the speaker diarization problem, the proposed approach takes into account the uncertainty associated with the explicit estimation of the speaker factors. The proposed decision criteria, based on Bayesian theory, was shown to generally outperform their non- Bayesian counterparts.

Accelerometer-based fetal movement detection

Monitoring fetal wellbeing is a compelling problem in modern obstetrics. Clinicians have become increasingly aware of the link between fetal activity (movement), well-being, and later developmental outcome. We have recently developed an ambulatory accelerometer-based fetal activity monitor (AFAM) to record 24-hour fetal movement. Using this system, we aim at developing signal processing methods to automatically detect and quantitatively characterize fetal movements. The first step in this direction is to test the performance of the accelerometer in detecting fetal movement against real-time ultrasound imaging (taken as the gold standard). This paper reports first results of this performance analysis.

SERS based detection of barcoded gold nanoparticle assemblies from within animal tissue

We have explored the potential of deep Raman spectroscopy, specifically surface enhanced spatially offset Raman spectroscopy (SESORS), for non-invasive detection from within animal tissue, by employing SERS-barcoded nanoparticle (NP) assemblies as the diagnostic agent. This concept has been experimentally verified in a clinic-relevant backscattered Raman system with an excitation line of 785 nm under ex vivo conditions. We have shown that our SORS system, with a fixed offset of 2-3 mm, offered sensitive probing of injected QTH-barcoded NP assemblies through animal tissue containing both protein and lipid. In comparison to that of non-aggregated SERS-barcoded gold NPs, we have demonstrated that the tailored SERS-barcoded aggregated NP assemblies have significantly higher detection sensitivity. We report that these NP assemblies can be readily detected at depths of 7-8 mm from within animal proteinaceous tissue with high signal-to-noise (S/N) ratio. In addition they could also be detected from beneath 1-2 mm of animal tissue with high lipid content, which generally poses a challenge due to high absorption of lipids in the near-infrared region. We have also shown that the signal intensity and S/N ratio at a particular depth is a function of the SERS tag concentration used and that our SORS system has a QTH detection limit of 10-6 M. Higher detection depths may possibly be obtained with optimization of the NP assemblies, along with improvements in the instrumentation. Such NP assemblies offer prospects for in vivo, non-invasive detection of tumours along with scope for incorporation of drugs and their targeted and controlled release at tumour sites. These diagnostic agents combined with drug delivery systems could serve as a “theranostic agent”, an integration of diagnostics and therapeutics into a single platform.

IS-RT-PCR assay detection of MT-MMP in a human breast cancer cell line

The in situ-reverse transcription-polymerase chain reaction (IS-RT-PCR) is a method that allows the direct localisation of gene expression. The method utilises the dual buffer mediated activity of the enzyme rTth DNA polymerase enabling both reverse transcription and DNA amplification. Labelled nucleoside triphosphates allow the site of expression to be labelled, rather than the PCR primers themselves, giving a more accurate localisation of transcript expression and decreased background than standard in situ hybridisation (ISH) assays. The MDA-MB-231 human breast carcinoma (HBC) cell line was assayed via the IS-RT-PCR technique, using primers encoding MT-MMP (membrane-type matrix metalloproteinase) and human β-actin. Our results clearly indicate baseline expression of MT-MMP in the relatively invasive MDA-MB-231 cell line at a signal intensity similar to the housekeeping gene β-actin, and results following induction with Concanavalin A (Con A) are consistent with our previous results obtained via Northern blotting.