Distributed low-power image processing in wireless sensor networks for intelligent video surveillance applications

Distributed Wireless Smart Camera (DWSC) network is a special type of Wireless Sensor Network (WSN) that processes captured images in a distributed manner. While image processing on DWSCs sees a great potential for growth, with its applications possessing a vast practical application domain such as security surveillance and health care, it suffers from tremendous constraints. In addition to the limitations of conventional WSNs, image processing on DWSCs requires more computational power, bandwidth and energy that presents significant challenges for large scale deployments. This dissertation has developed a number of algorithms that are highly scalable, portable, energy efficient and performance efficient, with considerations of practical constraints imposed by the hardware and the nature of WSN. More specifically, these algorithms tackle the problems of multi-object tracking and localisation in distributed wireless smart camera net- works and optimal camera configuration determination. Addressing the first problem of multi-object tracking and localisation requires solving a large array of sub-problems. The sub-problems that are discussed in this dissertation are calibration of internal parameters, multi-camera calibration for localisation and object handover for tracking. These topics have been covered extensively in computer vision literatures, however new algorithms must be invented to accommodate the various constraints introduced and required by the DWSC platform. A technique has been developed for the automatic calibration of low-cost cameras which are assumed to be restricted in their freedom of movement to either pan or tilt movements. Camera internal parameters, including focal length, principal point, lens distortion parameter and the angle and axis of rotation, can be recovered from a minimum set of two images of the camera, provided that the axis of rotation between the two images goes through the camera's optical centre and is parallel to either the vertical (panning) or horizontal (tilting) axis of the image. For object localisation, a novel approach has been developed for the calibration of a network of non-overlapping DWSCs in terms of their ground plane homographies, which can then be used for localising objects. In the proposed approach, a robot travels through the camera network while updating its position in a global coordinate frame, which it broadcasts to the cameras. The cameras use this, along with the image plane location of the robot, to compute a mapping from their image planes to the global coordinate frame. This is combined with an occupancy map generated by the robot during the mapping process to localised objects moving within the network. In addition, to deal with the problem of object handover between DWSCs of non-overlapping fields of view, a highly-scalable, distributed protocol has been designed. Cameras that follow the proposed protocol transmit object descriptions to a selected set of neighbours that are determined using a predictive forwarding strategy. The received descriptions are then matched at the subsequent camera on the object's path using a probability maximisation process with locally generated descriptions. The second problem of camera placement emerges naturally when these pervasive devices are put into real use. The locations, orientations, lens types etc. of the cameras must be chosen in a way that the utility of the network is maximised (e.g. maximum coverage) while user requirements are met. To deal with this, a statistical formulation of the problem of determining optimal camera configurations has been introduced and a Trans-Dimensional Simulated Annealing (TDSA) algorithm has been proposed to effectively solve the problem.

Vision-based estimation of airborne target pseudobearing rate using hidden Markov model filters

The problem of estimating pseudobearing rate information of an airborne target based on measurements from a vision sensor is considered. Novel image speed and heading angle estimators are presented that exploit image morphology, hidden Markov model (HMM) filtering, and relative entropy rate (RER) concepts to allow pseudobearing rate information to be determined before (or whilst) the target track is being estimated from vision information.

Agent-based model of passenger flows in airport terminals

Passenger flow studies in airport terminals have shown consistent statistical relationships between airport spatial layout and pedestrian movement, facilitating prediction of movement from terminal designs. However, these studies are done at an aggregate level and do not incorporate how individual passengers make decisions at a microscopic level. Therefore, they do not explain the formation of complex movement flows. In addition, existing models mostly focus on standard airport processing procedures such as immigration and security, but seldom consider discretionary activities of passengers, and thus are not able to truly describe the full range of passenger flows within airport terminals. As the route-choice decision-making of passengers involves many uncertain factors within the airport terminals, the mechanisms to fulfill the capacity of managing the route-choice have proven difficult to acquire and quantify. Could the study of cognitive factors of passengers (i.e. human mental preferences of deciding which on-airport facility to use) be useful to tackle these issues? Assuming the movement in virtual simulated environments can be analogous to movement in real environments, passenger behaviour dynamics can be similar to those generated in virtual experiments. Three levels of dynamics have been devised for motion control: the localised field, tactical level, and strategic level. A localised field refers to basic motion capabilities, such as walking speed, direction and avoidance of obstacles. The other two fields represent cognitive route-choice decision-making. This research views passenger flow problems via a "bottom-up approach", regarding individual passengers as independent intelligent agents who can behave autonomously and are able to interact with others and the ambient environment. In this regard, passenger flow formation becomes an emergent phenomenon of large numbers of passengers interacting with others. In the thesis, first, the passenger flow in airport terminals was investigated. Discretionary activities of passengers were integrated with standard processing procedures in the research. The localised field for passenger motion dynamics was constructed by a devised force-based model. Next, advanced traits of passengers (such as their desire to shop, their comfort with technology and their willingness to ask for assistance) were formulated to facilitate tactical route-choice decision-making. The traits consist of quantified measures of mental preferences of passengers when they travel through airport terminals. Each category of the traits indicates a decision which passengers may take. They were inferred through a Bayesian network model by analysing the probabilities based on currently available data. Route-choice decision-making was finalised by calculating corresponding utility results based on those probabilities observed. Three sorts of simulation outcomes were generated: namely, queuing length before checkpoints, average dwell time of passengers at service facilities, and instantaneous space utilisation. Queuing length reflects the number of passengers who are in a queue. Long queues no doubt cause significant delay in processing procedures. The dwell time of each passenger agent at the service facilities were recorded. The overall dwell time of passenger agents at typical facility areas were analysed so as to demonstrate portions of utilisation in the temporal aspect. For the spatial aspect, the number of passenger agents who were dwelling within specific terminal areas can be used to estimate service rates. All outcomes demonstrated specific results by typical simulated passenger flows. They directly reflect terminal capacity. The simulation results strongly suggest that integrating discretionary activities of passengers makes the passenger flows more intuitive, observing probabilities of mental preferences by inferring advanced traits make up an approach capable of carrying out tactical route-choice decision-making. On the whole, the research studied passenger flows in airport terminals by an agent-based model, which investigated individual characteristics of passengers and their impact on psychological route-choice decisions of passengers. Finally, intuitive passenger flows in airport terminals were able to be realised in simulation.

A computer image analysis system for microvessel density measurement in solid tumours

Microvessel density (MVD) is a widely used surrogate measure of angiogenesis in pathological specimens and tumour models. Measurement of MVD can be achieved by several methods. Automation of counting methods aims to increase the speed, reliability and reproducibility of these techniques. The image analysis system described here enables MVD measurement to be carried out with minimal expense in any reasonably equipped pathology department or laboratory. It is demonstrated that the system translates easily between tumour types which are suitably stained with minimal calibration. The aim of this paper is to offer this technique to a wider field of researchers in angiogenesis.

Agreement between temporal and spatial gait parameters from an instrumented walkway and treadmill system at matched walking speed

Background Commercially available instrumented treadmill systems that provide continuous measures of temporospatial gait parameters have recently become available for clinical gait analysis. This study evaluated the level of agreement between temporospatial gait parameters derived from a new instrumented treadmill, which incorporated a capacitance-based pressure array, with those measured by a conventional instrumented walkway (criterion standard). Methods Temporospatial gait parameters were estimated from 39 healthy adults while walking over an instrumented walkway (GAITRite®) and instrumented treadmill system (Zebris) at matched speed. Differences in temporospatial parameters derived from the two systems were evaluated using repeated measures ANOVA models. Pearson-product-moment correlations were used to investigate relationships between variables measured by each system. Agreement was assessed by calculating the bias and 95% limits of agreement. Results All temporospatial parameters measured via the instrumented walkway were significantly different from those obtained from the instrumented treadmill (P < .01). Temporospatial parameters derived from the two systems were highly correlated (r, 0.79–0.95). The 95% limits of agreement for temporal parameters were typically less than ±2% of gait cycle duration. However, 95% limits of agreement for spatial measures were as much as ±5 cm. Conclusions Differences in temporospatial parameters between systems were small but statistically significant and of similar magnitude to changes reported between shod and unshod gait in healthy young adults. Temporospatial parameters derived from an instrumented treadmill, therefore, are not representative of those obtained from an instrumented walkway and should not be interpreted with reference to literature on overground walking.

Optimal camera planning under versatile user constraints in multi-camera image processing systems

The selection of optimal camera configurations (camera locations, orientations, etc.) for multi-camera networks remains an unsolved problem. Previous approaches largely focus on proposing various objective functions to achieve different tasks. Most of them, however, do not generalize well to large scale networks. To tackle this, we propose a statistical framework of the problem as well as propose a trans-dimensional simulated annealing algorithm to effectively deal with it. We compare our approach with a state-of-the-art method based on binary integer programming (BIP) and show that our approach offers similar performance on small scale problems. However, we also demonstrate the capability of our approach in dealing with large scale problems and show that our approach produces better results than two alternative heuristics designed to deal with the scalability issue of BIP. Last, we show the versatility of our approach using a number of specific scenarios.

The velocity synchronous discrete Fourier transform for order tracking in the field of rotating machinery

Diagnostics of rotating machinery has developed significantly in the last decades, and industrial applications are spreading in different sectors. Most applications are characterized by varying velocities of the shaft and in many cases transients are the most critical to monitor. In these variable speed conditions, fault symptoms are clearer in the angular/order domains than in the common time/frequency ones. In the past, this issue was often solved by synchronously sampling data by means of phase locked circuits governing the acquisition; however, thanks to the spread of cheap and powerful microprocessors, this procedure is nowadays rarer; sampling is usually performed at constant time intervals, and the conversion to the order domain is made by means of digital signal processing techniques. In the last decades different algorithms have been proposed for the extraction of an order spectrum from a signal sampled asynchronously with respect to the shaft rotational velocity; many of them (the so called computed order tracking family) use interpolation techniques to resample the signal at constant angular increments, followed by a common discrete Fourier transform to shift from the angular domain to the order domain. A less exploited family of techniques shifts directly from the time domain to the order spectrum, by means of modified Fourier transforms. This paper proposes a new transform, named velocity synchronous discrete Fourier transform, which takes advantage of the instantaneous velocity to improve the quality of its result, reaching performances that can challenge the computed order tracking.

A new procedure for using envelope analysis for rolling element bearing diagnostics in variable operating conditions

In the field of rolling element bearing diagnostics envelope analysis, and in particular the squared envelope spectrum, have gained in the last years a leading role among the different digital signal processing techniques. The original constraint of constant operating speed has been relaxed thanks to the combination of this technique with the computed order tracking, able to resample signals at constant angular increments. In this way, the field of application of squared envelope spectrum has been extended to cases in which small speed fluctuations occur, maintaining the effectiveness and efficiency that characterize this successful technique. However, the constraint on speed has to be removed completely, making envelope analysis suitable also for speed and load transients, to implement an algorithm valid for all the industrial application. In fact, in many applications, the coincidence of high bearing loads, and therefore high diagnostic capability, with acceleration-deceleration phases represents a further incentive in this direction. This paper is aimed at providing and testing a procedure for the application of envelope analysis to speed transients. The effect of load variation on the proposed technique will be also qualitatively addressed.

Fault symptoms of rolling element bearings under variable operating conditions - a multi domain analysis

Diagnostics of rolling element bearings have been traditionally developed for constant operating conditions, and sophisticated techniques, like Spectral Kurtosis or Envelope Analysis, have proven their effectiveness by means of experimental tests, mainly conducted in small-scale laboratory test-rigs. Algorithms have been developed for the digital signal processing of data collected at constant speed and bearing load, with a few exceptions, allowing only small fluctuations of these quantities. Owing to the spreading of condition based maintenance in many industrial fields, in the last years a need for more flexible algorithms emerged, asking for compatibility with highly variable operating conditions, such as acceleration/deceleration transients. This paper analyzes the problems related with significant speed and load variability, discussing in detail the effect that they have on bearing damage symptoms, and propose solutions to adapt existing algorithms to cope with this new challenge. In particular, the paper will i) discuss the implication of variable speed on the applicability of diagnostic techniques, ii) address quantitatively the effects of load on the characteristic frequencies of damaged bearings and iii) finally present a new approach for bearing diagnostics in variable conditions, based on envelope analysis. The research is based on experimental data obtained by using artificially damaged bearings installed on a full scale test-rig, equipped with actual train traction system and reproducing the operation on a real track, including all the environmental noise, owing to track irregularity and electrical disturbances of such a harsh application.

Bearing Fault Diagnostics Using the Spectral Pattern Recognition

In the field of diagnostics of rolling element bearings, the development of sophisticated techniques, such as Spectral Kurtosis and 2nd Order Cyclostationarity, extended the capability of expert users to identify not only the presence, but also the location of the damage in the bearing. Most of the signal-analysis methods, as the ones previously mentioned, result in a spectrum-like diagram that presents line frequencies or peaks in the neighbourhood of some theoretical characteristic frequencies, in case of damage. These frequencies depend only on damage position, bearing geometry and rotational speed. The major improvement in this field would be the development of algorithms with high degree of automation. This paper aims at this important objective, by discussing for the first time how these peaks can draw away from the theoretical expected frequencies as a function of different working conditions, i.e. speed, torque and lubrication. After providing a brief description of the peak-patterns associated with each type of damage, this paper shows the typical magnitudes of the deviations from the theoretical expected frequencies. The last part of the study presents some remarks about increasing the reliability of the automatic algorithm. The research is based on experimental data obtained by using artificially damaged bearings installed in a gearbox.

Diagnostic of Rolling Element Bearings with Envelope Analysis in Non-Stationary Conditions

In the field of rolling element bearing diagnostics, envelope analysis has gained in the last years a leading role among the different digital signal processing techniques. The original constraint of constant operating speed has been relaxed thanks to the combination of this technique with the computed order tracking, able to resample signals at constant angular increments. In this way, the field of application of this technique has been extended to cases in which small speed fluctuations occur, maintaining high effectiveness and efficiency. In order to make this algorithm suitable to all industrial applications, the constraint on speed has to be removed completely. In fact, in many applications, the coincidence of high bearing loads, and therefore high diagnostic capability, with acceleration-deceleration phases represents a further incentive in this direction. This chapter presents a procedure for the application of envelope analysis to speed transients. The effect of load variation on the proposed technique will be also qualitatively addressed.

An analysis of regression models for predicting the speed of a wave glider autonomous surface vehicle

An important aspect of robotic path planning for is ensuring that the vehicle is in the best location to collect the data necessary for the problem at hand. Given that features of interest are dynamic and move with oceanic currents, vehicle speed is an important factor in any planning exercises to ensure vehicles are at the right place at the right time. Here, we examine different Gaussian process models to find a suitable predictive kinematic model that enable the speed of an underactuated, autonomous surface vehicle to be accurately predicted given a set of input environmental parameters.

Point-to-point speed enforcement : recommendations for better practice

Point-to-point speed cameras are a relatively new and innovative technological approach to speed enforcement that is increasingly been used in a number of highly motorised countries. Previous research has provided evidence of the positive impact of this approach on vehicle speeds and crash rates, as well as additional traffic related outcomes such as vehicle emissions and traffic flow. This paper reports on the conclusions and recommendations of a large-scale project involving extensive consultation with international and domestic (Australian) stakeholders to explore the technological, operational, and legislative characteristics associated with the technology. More specifically, this paper provides a number of recommendations for better practice regarding the implementation of point-to-point speed enforcement in the Australian and New Zealand context. The broader implications of the research, as well as directions for future research, are also discussed.