Master of Philosophy (M.Phil.)

Permanent URI for this collectionhttp://192.248.9.226/handle/123/18724

Browse

Now showing 1 - 20 of 27

item: Thesis-Abstract
A Novel stair-climbing mobile platform and its motion control system
(2022) Thamel WSR; Munasinghe R
Independent life is one of the most signiﬁcant concerns for a disabled person specially when considering their mobility requirements. Due to the availability of facilities, both indoor and outdoor, wheelchair users face many diﬃculties when entering a building via staircases. Particularly, in the developing world, the concern about diﬃculties faced by the disabled and elderly community in accessing buildings is inadequate. As a result, most of the wheelchair users are forced to stay home and it makes their lives dependent. Since many years, scientists and researchers have been working for providing eﬀective solutions for the mobility issues of disabled community. Stair-climbing wheelchair is one of the most powerful and popular solutions which is currently in use for mobility requirements of the disabled community. But most of the existing stair-climbing wheelchair designs have a drawback of the lack of adaptability especially at the top and the bottom of the staircase As a solution, a novel stair-climbing mobile platform with improved stability and safety is presented in this thesis. In this research, the novel mobility platform has been designed referring to the standards of staircases which have been accepted worldwide. Based on those stipulated parameters, the maximum operating angle of this design is selected as forty degrees, which is the highest angle as per the standards. Based on the literature review, track based arrangement is used as the basic mechanism for the mobile platform design because of speed, grip and user comfort. By considering the main problem of the track mechanism, variable conﬁguration of track mechanism is used for the design in order to achieve adaptable arrangement to the staircase. Also detail analysis of design safety based on diﬀerent postures on staircases are discussed and basic conditions for stair-climbing operation are derived. By considering the prototype construction, a detailed description of development stages of prototype is explained based on practical approaches in order to minimize the defects in the ﬁnal product. In order to validate the stable and safe operation, simulation and experimental studies have been carried out. According to proposed novel motion plans for stair-climbing operation, the simulation study illustrates that there does not exist any sudden disturbances on the user’s operating path over staircase. By following the same procedure as used in simulation study, experimental studies have been carried out by implementing motion plans for each stage of operation. Based on the results, it conﬁrms that the operating path of the design follows a similar behaviour as discussed in the simulation study. For further justiﬁcation of the safety of the design, the actual operating path is well observed by using a video tracker software. Based on the actual operating path of the design, it conﬁrms that throughout the staircase, the design maintains a stable operating path without creating any sudden disturbances on the mobile platform. Thus, it conﬁrms that this design can provide a comfortable operation for the user whilst maintaining stable and safe posture throughout the operation.
item: Thesis-Abstract
Hydrogel based insoles for diabetic feet
(2022) Udayanandana RMRC; Silva P; Gunasekara T
This research considered an extensive study of double network poly (acrylic acid) and single network polyacrylamide hydrogel as an insole material for the diabetic foot by testing on compression strength, stress relaxation, compression fatigue, shear stress, and shock absorption properties. The previous studies on hydrogel fatigue were mainly focused on fatigue fracture in tension and crack propagation. This study focuses on the mechanical behaviour of hydrogels under diabetic foot-specific loading conditions. The expected testing conditions include minimum 20 000 fatigue cycles under maximum compression stress of 250 kPa at a 200 mm/ min strain rate. The hydrogel synthesis and testing started with double network poly (acrylic acid) hydrogel. The developed double network poly (acrylic acid) hydrogel displayed fatigue properties up to 3000 loading cycles at maximum stress of 390±30 kPa. Further, maximum average shear stress and shear modulus of 80 kPa and 140 kPa respectively were observed at 84% strain before fracture. Developed Single network polyacrylamide hydrogel displayed good fatigue properties up to 13,000 loading cycles at maximum stress of 520±50 kPa and 200 mm/min crosshead speed. When the maximum stress condition was reduced to 350±50 kPa, the maximum number of loading-unloading cycles was increased up to 20 000 indicating a single network polyacrylamide hydrogel capable of withstanding more than 20 000 cycles at 250 kPa. Hydrogels showed superior recoverable and viscoelastic properties when compared with available insole materials. The developed finite element model was validated with pressure insole test data and used to investigate the pressure distribution properties and to optimize the thickness suitable for insole applications The additional properties of a hydrogel such as high thermal capacity and structural similarity to soft tissues are seen as added advantages when compared to other insole materials to prevent re-ulceration.
item: Thesis-Abstract
Implementation of a pedobarographic system through effective crosstalk suppression with an all programmable system on chip
(2022) Warnakulasuriya AS; De Silva A; Jayasinghe S
Diabetes Mellitus which is characterized by longstanding hyperglycemia, promote diabetic peripheral neuropathy (DPN) and peripheral artery diseases (PAD) which invoke ulcerations especially on the foot plantar ultimately leads to amputations related morbidity and mortality. Recent clinical studies have identified distinct associations between DPN and foot plantar pressure, and PAD and foot plantar temperature. Hence, development of technology to analyze foot plantar pressure (pedobarography) and foot plantar temperature (pedothermography) to infer DPN and PAD associated diseases at the onset have recently stirred a significant interest among the scientific community. In the present study, we have primarily investigated the possibility of implementing a highly accurate large piezoresistive platform sensor array (a pedobrographic system) with an improved readout mechanism. We devised a readout circuit to mitigate inherent crosstalk interference with an improved scanning architecture implemented using a decoder-transistor based row driving electrodes and related electronics compatible with high frequency sensing. Then, the developed readout circuit was extensively validated and the proposed implementation was able to make measurements of significant accuracy with errors < 1% while not compromising the shape accuracy of the measured object. Initiatives were also taken to improve the data acquisition rate despite massive sensor data influx from 30,000 sensels. A Xilinx Zynq APSoC based data acquisition system was implemented to scan the entire array with 30,000 sensels and the analysis showed that the system demonstrated expected behavior. Overall, the proposed implementation entertained both static and dynamic pressure measurements of a foot plantar. A subsequent static calibration of the piezoresisitive sensor array was conducted using weight plates and the calibrated sensor array was validated against an existing commercial plantar pressure measurement system to determine its performance. In addition, possibility of implementing a screening tool for pedothermographic assessment using near infrared (NIR) technology was investigated to provide an overall assessment of both foot planters in a single frame and record both regional and point temperatures of the foot plantar. The proposed thermal imaging system is anticipated to be used in routine clinical assessment of diabetic foot complications at diabetic clinics to provide improved diagnostics, thereby contributing to prevention of diabetic foot ulcerations, amputations and related morbidity.
item: Thesis-Full-text
Performance analysis of the power splitting simultaneous lightwave information and power transfer (PS-SLIPT) architecture
(2021) Morapitiya SS; Jayakody DNK; Udayanga LWPR
Recent studies done on Simultaneous Lightwave Information and Power Transfer (SLIPT) has become a hot topic among the research community. The importance of the SLIPT is to harvest energy using light sources while decoding the information. In this thesis work, we present the mathematical framework for the Power Splitting (PS) based SLIPT system and study the performance of the PS-SLIPT and Time Splitting (TS)-SLIPT architectures. Moreover, we quantitatively study the harvested energy with different Field of View (FoV) angles of the Light Emitting Diode (LED) and the Photodiode (PD). In addition, analyze the important parameter of the Visible Light Communication (VLC) system to achieve maximum received power and we consider the amount of harvested energy for different Direct Current (DC) values. Overall, concludes that the Field of View (FoV) and DC bias signals are directly affected by SLIPT systems. Using numerical simulations, we demonstrate the performance of the both architectures to enhance the QoS of information decoding data rate, amount of harvested energy and trustworthiness of the information. Further, our research work extend to Simultaneous Wireless Information and Power Transfer (SWIPT) technique is introduced in Radio Frequency (RF) communication to carry both information and power in same medium. In this approach, the energy can be harvested while decoding the information carries in an RF wave. Recently, the same concept apply in VLC namely SLIPT, which is highly recommended in an indoor applications to overcome the problem facing in RF communication. Thus, the SLIPT is introduce to transmit the power through a Light Emitting Diode (LED) luminaries. In this work, we compare both SWIPT and SLIPT technologies and realize SLIPT technology archives increase performance in terms of the amount of harvested energy, outage probability and error rate performance.
item: Thesis-Abstract
Modeling of polarization insensitive phase sensitive amplifier for phase regeneration
(2021) Munasinghe DR; Ruwan LWP
This thesis describes a novel conﬁguration which implements a polarization insensitive phase sensitive ﬁber optic parametric ampliﬁer for phase regeneration. The proposed design can be used to address the inherent gain degradation issue in a polarization insensitive phase sensitive ampliﬁer when polarization diversity loop is incorporated. This research investigates the possibilities for a signiﬁcant gain enhancement when the polarization diversity loop is implemented. At ﬁrst, a baseline for a phase sensitive ampliﬁer scheme is developed by taking diﬀerent parameters into consideration. In this case, an extensive characterization is carried out to identify the variations of phase sensitive ﬁber ampliﬁer gain associated with the parameter changes. Once the baseline is developed, the polarization diversity loop is implemented in the phase sensitive ampliﬁer in order to make the phase sensitive ﬁber ampliﬁer insensitive to polarization state variations of the input signals. Moreover, the existing issues in the polarization diversity loop are identiﬁed. Finally, a strategic design is developed to overcome one of the major issues that persists in the existing polarization diversity loop conﬁgurations. It is the degradation of maximum achievable gain due to not being able to utilize the total pump power for parametric ampliﬁcation process as only half of the pump power is available after polarization splitting at the input of the loop. The proposed design includes the concatenation of ﬁber pieces together to explore the possibility of gain enhancement of the diversity loop. To assess the performance of the proposed design, an extensive analysis is carried out. The gain enhancement introduced to the existing system is presented and suggestions are made for further improvements in the future.
item: Thesis-Abstract
3-Dimensional spatial channel model for multi-storeyed indoor environments
(2021) Tennakoon TMP; Wavegedara KCB
A three-dimensional (3-D) geometry-based stochastic model (GBSM) is presented for various types of multi-story indoor environments. The proposed model assumed that the scatterers are distributed within a spheroid, where the transmitter and the receiver are located at the focal points of the spheroid. The proposed model provides the probability density functions (PDFs) of the angle of arrival (AoA), the time of arrival (ToA) and the spatial correlation coe cients correspondence with several channel parameters of the channel. By considering non-uniform scatterer distributions, the spheroid GBSM is extended for multistory indoor environments. Closed-form expressions are derived for the joint and marginal PDFs of the AoA in both the elevation and azimuth planes and the ToA. The analytically-derived PDFs of the AoA and ToA obtained for Gaussian and Rayleigh scatterer distributions are compared against those obtained from the ray-tracing simulation of typical indoor environments. The standard deviation values of Gaussian and Rayleigh scatterer distributions are chosen to provide the best possible approximation to the PDFs of the AoA and the ToA obtained from simulation. Our results clearly indicate that the analyticallyderived PDFs of the AOA and the TOA for Gaussian and Rayleigh scatterer distributions are in much closer agreement with those obtained from ray-tracing simulation than for uniform scatterer distribution. However, analytically-derived PDFs of the AOA and the TOA for Gaussian scatterer distribution show closest agreement with those PDFs obtained from the simulations. A generalized 3D channel model with an arbitrator scatterer distribution point is proposed based on the spheroid GBMS. The proposed channel model is assumed that the scatterers to be distributed according to the Gaussian distribution about an arbitrary point within the spheroid. Closed-form expressions are derived for the joint PDFs of the AoA, marginal PDFs in both the elevation and azimuth planes, as well as for the marginal PDF of the ToA. Numerical results are utilized for the veri cation of the derived-closed form mathematical expressions. More-over, the obtained marginal PDFs of AoA and TOA are compared against PDFs obtained from the simulation of an indoor environment using ray-tracing tool. By choosing a proper scatterer distribution center point based on the actual indoor propagation environment and a suitable value for the standard deviation of the scatterer region, the proposed 3-D model of the channel can be exploit the performance of the wireless communication technologies and systems in indoor environments. The spheroid GBSM is extended to a 3D geometry-based spatial correlation model for multiple-input multiple-output (MIMO) communication environments. Approximated closed-form expressions are obtained for the normalized spatial correlation coe cients of frequency non-selective Rician fading channels. As a special case, the normalized spatial coe cients are derived for Gaussian scatterer distribution. Closed-form expressions developed are veri ed by the simulation results obtained using the WINNER Phase II channel model (WIM2). Furthermore, the capacity performance of MIMO channels is investigated using the proposed geometry-based correlation model. Our results have clearly demonstrated that the proposed 3D spatial correlation model can be used to investigate the performance of the frequency non-selective Rician or Rayleigh fading MIMO channels with di erent antenna con gurations accurately.
item: Thesis-Abstract
Design and characterizing of optical carrier recovery scheme
(2021) Alahakoon SE; Udayanga LWPR
The field of optical communication has gained rapid development throughout the past years. With the exponentially increasing demand for the internet, the existing transmission bandwidth of fiber optic communication has become insufficient to meet up with end-user requirements. Hence, it is necessary to increase the data rate while enhancing the quality of optical communication systems. Due to the unavailability of high-speed electronics, typical intensity modulation is hard to improve further. One future proof option is to use phase modulated schemes or a combination of phase and amplitude modulation schemes. In this scenario, simultaneous amplitude and phase regeneration schemes are implemented thus making phase synchronization a major concern during the regeneration process. This research is intended to extract the carrier signal phase using the principle of four wave mixing (FWM) and to implement a successful optical carrier recovery scheme. The noise influences from the signal laser for the recovered carrier have been investigated both theoretically and experimentally. Also, the dispersion effects have been analyzed experimentally
item: Thesis-Abstract
Optical Injection locking and characterization for optical communication
(2020) Samarakkody GS; Udayanga LWP.R; Kumarayapa YA
The rise of demand for mobile and broadband communication has caused a worldwide requirement of sophisticated transmitters. High volume and low-cost manufacturing of semiconductor optoelectronic devices are playing an integral role in allowing for the deployment of high-speed communication links. Requirements of side mode suppression, reducing nonlinearities, reducing relative intensity noises and chirps, enhancing the bandwidth are key parameters that are leading for the enhancement of quality and the performance of fiber optic communication system. Optical injection locking (OIL) is one technique to overcome these limitations arising in developing better optical transmitters. Improved device performances can be achieved by locking the high-quality phase and frequency of the master laser to a low-quality slave laser. In this research, a proper optical injection locking system will be developed and characterized by the perspective of improving certain parameters in optical communication systems. A simulation model will be developed and then the characterization will be done for the locking range, bandwidth, amplitude, and phase noise transfer function. The anticipated output will be helpful for several applications like developing optical synchronization in phase-sensitive amplifiers which is useful for improving the transparent length and the operating margins of the optical network system. Also, it will build up a pathway to increase the speed of the optical communication system by the increment of the bandwidth. Index terms— Optical injection locking, locking range, bandwidth
item: Thesis-Full-text
Through hole soldering system with automatic optical inspection
(2020) Fonseka CLSC; Jayasinghe JAKS
medical, manufacturing, agricultural, surveillance etc. Such applications consist of numerous algorithms to fulfil the ultimate requirement of the end product. The role of automatic vision for classifying the quality of solder joints in the Electronics Manufacturing Services (EMS) industry has become significant because the increasing cost of labour, skill dependency, attitudes, time variance of human operators have narrowed down the quality of their service. Several major companies develop stand-alone Automatic Optical Inspection (AOI) systems with proprietary algorithms that contains multiple cameras operating inside a specific lighting environment for the inspection of Surface Mount Devices (SMD) and Through-Hole Technology (THT) components. However, the accuracy and maturity level for the automatic inspection of the quality of solder joints have yet to reach its ultimate goal due to the complex appearance of solder joint surface. Even though, many AOI systems are available, no vision system has been developed to integrate with a soldering robotic system to provide quality classification of solder joints in real time without prior teaching of individual joints and enclosed chambers with specific lighting to operate. In this research, a THT soldering system with an AOI and Computer Aided Design (CAD) tool has been developed to classify the quality of THT solder joints without prior teaching and specific lighting requirements. The design consists of three main stages:  AOI system for quality classification of THT solder joints (Major Research)  CAD tool for extracting physical parameters of each component (Minor Research)  THT Soldering system to perform automatic soldering (Minor Research) The AOI system mainly operates in three stages. The first stage involves with precise alignment of the Printed Circuit Board (PCB) to the origin of the THT soldering system using feature extraction and template matching techniques. This approach provide a reliable and robust PCB verification and alignment capability compared to Hough transform based alignment method proposed by Nang Seng Siri Mar. Furthermore, this methodology rendered a better outcome even in distinguishing between vias, pads and test-points in addition to conventional fiducial markings. The identification of the solder pad and the verification of the component availability is performed prior to soldering in the second stage. The automatic identification of solder pad regardless of its plated colour and surrounding solder mask colour made a significant improvement over the methodology proposed by T. Y. Ong, Z. Samad and M. M. Ratnam, based on prior teaching of individual pads using artificial neural networks. In addition to that, the implemented methodology provides online positioning accuracy calibration which is not available in any commercial soldering robotic systems. The component availability is assured by precisely segmenting the component lead top from the identified solder pad. The developed algorithms could render a better outcome even for component leads which possess a minimum colour dissimilarity with their surrounding drill-hole region. No commercial soldering robotic system is capable of verifying component availability prior to soldering. Further, the segmented lead is used to establish a relationship between prior and post soldering stages of the solder pad in order to provide a robust solder quality classification capability. The classification of the soldering quality for short circuit (solder bridging), voids inside the drill-hole, voids on solder pad and excess solder, is carried out after soldering during the third stage. The implemented algorithms could render an improved recognition rate even with applied flux, illuminated pad regions, uneven illumination distribution and shadows on the solder joint surface. Such a widely distributed quality inspection capability is iii not covered in any of the reported studies. Neither commercial AOI systems nor soldering robotic systems perform real time inspection of soldering quality just after the soldering of THT components. Moreover, the precise localization of defective areas inside the solder joint, enables the robotic system to perform automatic reworking on defective solder joints adhering to IPC regulations with minimum user interaction. Such automatic reworking capability is not available with any commercial soldering robotic system in the market today. The CAD system extracts the geometrical information of components and their pads such as the component location, its orientation, size of the solder pad and drill hole, height of nearby components, the width of the connected PCB track to the solder pad from the respective CAD file and visualize the 2D view of the PCB to the user in a Graphical User Interface (GUI). This information enhances the intelligence and the situational awareness of both robotic system and AOI. The THT soldering system is a four-axis robotic platform that performs soldering on selected solder pads through the CAD system GUI. Its operation is mainly controlled by the vision system and the information acquired from the CAD system. The implemented THT soldering system together with integrated AOI and CAD tool provides a new concept in the EMS industry by replacing the manual inspection of THT solder joints with automatic inspection and providing automatic rework capability on defective solder joints within a single platform. The performance of the complete system was evaluated under different illumination levels, flux residues, different types of component leads, colour combinations of solder pads and solder mask colours, wide variety of solder pad neighbourhoods and a range of solder pad sizes. Each stage of the AOI was able to provide a significant improvement over the reported studies and commercial systems. The automatic identification of solder pad and the verification of component availability could provide nearly a 98% of recognition rate for both cases. However, the existence of highly illuminated pad regions and overheated solder joints surfaces degrades the performance of the classification of not soldered regions by 2% and excess solder detection by 5% respectively (subjected to a sample size of 200 solder joints). Even though, a slight performance reduction is there due to such extreme conditions, the proposed approach provides an automated solution for soldering and quality assurance within a single platform while solving several problems in the reported studies and commercial systems effectively.
item: Thesis-Full-text
Development of organ stiffness models for haptic feedback in laparoscopic surgery simulation
Kannangara, SM; Dayananda, NWN; Kumarage, S
Laparoscopic surgery is the most common Minimally Invasive Surgery (MIS) performed routinely for certain procedures such as appendectomy and Cholecystectomy. Laparoscopic surgical procedures are very complex compared to open surgeries and require a higher level of experience and expertise. A comprehensive training session on surgical simulator handling for trainee surgeons is highly recommended before the hands-on training in a real surgery. Comprehensive surgery simulators such as physical phantoms which are available for training are expensive and not readily available in many health care centers around the world. VR simulators have a great potential to revalorize the training paradigm of surgical interns. The haptic feedback plays as equally as visual feedback to provide a realistic environment to trainees. Realistic organ-force model is a key requirement of a VR simulator to experience real-time tool-tissue interaction forces. However, modeling real tissue properties has not been achieved due to several limitations such as the inaccessibility to in-vivo tissue properties, the complex behavior of biological tissues and anatomical variability. We have adopted an alternative approach to incorporate force feedback to VR simulators. The abdomen organ models (liver, gallbladder, stomach, bone, and vessel) were generated using the color Cryosection dataset of the Visible Human Project. A novel method was applied to render forces by fine-tuning the stiffness of organ model and integrating the three force ranges: soft, mild/firm and hard into organ models using feedback received from expert surgeons. The proposed system provides the interaction forces through a haptic device with six Degrees of Freedom (DoF) position sensing and three DOF force feedback. The simulated organ models were evaluated by two experienced surgeons. The proposed haptic models were mostly in harmony with their experience in real-world tool-tissue interaction and the overall accuracy of identifying the correct organ property was more than 68%. The organ models were also tested with senior registrars. The results showed a considerable improvement amounting to more than 34% chances of selecting the correct organ property after training.
item: Thesis-Full-text
Activity recognition combined with scene context and action sequence
Ramasinghe, SC; Rodrigo, R
In this study, we investigate the problem of automatic action recognition and classification of videos. First, we present a convolutional neural network architecture, which takes both motion and static information as inputs in a single stream. We show the network is able to treat motion and static information as different feature maps and extract features off them, even though stacked together. By our results, we justify the use of optic flows as the raw information of motion. We demonstrate that our network is able to surpass state-of-the-art hand-engineered feature methods. Furthermore, the effect of providing static information to the network, in the task of action recognition, is also studied and compared here. Then, a novel pipeline is proposed, in order to recognize complex actions. A complex activity is a temporal composition of subevents, and a sub-event typically consists of several low level micro-actions, such as body movement, done by different actors. Extracting these micro actions explicitly is beneficial for complex activity recognition due to actor selectivity, higher discriminative power, and motion clutter suppression. Moreover, considering both static and motion features is vital for activity recognition. However, how to control the contribution from each feature domain optimally still remains uninvestigated. In this work, we extract motion features in micro level, preserving the actor identity, to later obtain a high-level motion descriptor using a probabilistic model. Furthermore, we propose two novel schemas for combining static and motion features: Cholesky transformation based and entropy based. The former allows to control the contribution ratio precisely, while the latter uses the optimal ratio mathematically. The ratio given by an entropy based method matches well with the experimental values obtained by a Choleksy transformation based method. This analysis also provides the ability to characterize a dataset, according to its richness in motion information. Finally, we study the effectiveness of modeling the temporal evolution of sub-event using an LSTM network. Experimental results demonstrate that the proposed technique outperforms state- of-the-art, when tested against two popular datasets.
item: Thesis-Full-text
Design and implementation of a multi-port power converter topology for DC nano-grid
(2019) Gunawardena LHPN; Nayanasiri MDR; Edussooriya C
This thesis presents a novel multi-port power converter topology for DC nanogrid applications. The proposed topology integrates energy sources, loads and energy storing elements using DC-link and magnetic coupling using a single converter. As a result, it has fewer component counts and conversion stages than the individual converters for each element in the nano-grid, which paves the way for a more e cient system. The rst part of this thesis presents a PV converter topology designed and developed in the laboratory. This circuit topology integrates two PV modules and boosts the input voltage into a 120V DC voltage level. However, switching loss of the converter is signi cant due to the hard-switching operation. Therefore, switching control strategy of the converter has been modi ed to minimize switching losses with the assistance of the existing parasitic elements. The operation of the power converter with the proposed switching control strategy is mathematically analyzed and veri ed using simulation results. The design is further validated using the experimental results obtained using a 250 W hardware prototype. Moreover, a bi-directional high step-up/down converter is designed and developed to integrate an energy storing element into the system. The bi-directional converter step downs 120 V DC link voltage to an extremely low voltage (10-16 V DC) to charge a Li-ion battery pack. When the solar power is not available, the proposed converter discharges the Li-ion battery to regulate the 120 V DC link. The operation of the battery interfacing converter is validated and veri ed using both simulation and experimental results. The conclusions and suggestions for the further development have been presented at the end of this thesis.
item: Thesis-Full-text
Four- dimensional sparse filters for near real-time light field processing
(2019) Premaratne IWASU; Edussooriya CUS
Light is a fundamental form of conveying information. Sensing of light through conventional cameras leads to images and videos. In contrast to conventional images and videos, which capture only the directional variation of the intensity of light rays emanating from a scene, light fields capture the spatial variation as well. This richness of information has been exploited to accomplish novel tasks that are not possible with conventional images and videos, such as post-capture digital refocusing and depth filtering. As a result of the massive data volume captured by a light field, the light field processing algorithms require higher memory and computational requirement. This is a major drawback for employing light fields in real-time applications. Hence, there is a need for investigating novel low-complexity light field processing algorithms that can be implemented in real-time applications. In this study, we address this critical research problem using multidimensional linear filter theory to develop novel low-complexity and sparse filters for light field processing. To this end, the work presented in this thesis focus on two major scenarios; light field denoising and volumetric refocusing. First, we present a novel low-complexity light field denoising algorithm, utilizing the sparsity of the region of support of a light field in the frequency domain. It turns out that the proposed filter runs in near real-time, compared to the previously reported light field denoising methods which take minutes. Next, a 4-D sparse filter for volumetric refocusing is presented. The proposed sparse filter provides 72% reduction of computational complexity compared to a non-sparse filter, with negligible distortion in fidelity.
item: Thesis-Full-text
Development of a customised protocol for diagnosis and treatment of obesity specific foot biomechanics
(2019) Kathirgamanathan B; Silva P; Fernandez J
The foot plays a crucial role in locomotion, impact bearing, and vertical stability and hence disruptions to the foot biomechanics can be detrimental to a person's ability to carry out daily activities. Foot disorders and pathologies can cause alterations in the overall foot and lower limb biomechanics. This thesis develops and applies a methodology using computational and experimental methods to analyse the lower limb biomechanics. The biomechanical analysis allowed foot pressure, stress and other changes that occur during gait due to pathology to be identi ed. The developed methods were implemented on obese subjects to characterise changes in the lower limb biomechanics caused due to increased loading following which some computational studies were carried out to further characterise the link between obesity and osteoarthritis. Finally, a subject speci c modelling method is explored in order to take the initial methodology which is purely a research solution into one that may be used in a clinical setting. The ndings of this study highlight the importance of studying the foot biomechanics as a whole whilst undertaking biomechanical studies, as the lower limb is a chain where problems occurring at the foot can be observed higher up in the lower limb and vice versa. The results of this study suggest that alterations in the foot posture are a key indicator of increased internal stress and pressure. The data collected for this thesis was from a South Asian (Sri Lankan) population, and hence a useful data-set for future comparison with the large body of European data currently available has been obtained.
item: Thesis-Full-text
Multi-Modal Evidence Filtering in Wireless Sensor Networks
Weeraddana, DM; Dias, SAD; Kulasekera, EC; Walgama, KS
A novel framework named Dempster-Shafer Information Filtering for in- formation processing in Distributed Sensor Networks (DSNs) is presented. More- over, distributed algorithms to implement spatio-temporal ltering applications in grid sensor networks are presented within the context of the framework. The framework facilitates processing multi-modality sensor data with a high noise level. Moreover, we compare intuitively appealing results against Dempster- Shafer fusion to grant further credence to the proposed framework. The concept of the proposed framework is based on the belief notions in Dempster-Shafer (DS) evidence theory. It enables one to directly process tem- porally and spatially distributed multi-modality sensor data to extract meaning buried in the noise clutter. Certain facts on lter parameter's selection impose several challenges in the design of the Information Filter. This is analysed using a re propagation scenario when high noise is present in the sensed data. Infor- mation bandwidth and the sluggishness of the lter are traded-o to minimise the e ect of the noise in the output evidence signal. From the application point of view, we address a Wireless Sensor Network (WSN) deployed in a multi-stoery building which can be e ectively used to convey information to relevant parties ( re ghters in their rescue operations) during an emergency situation. Therefore, a re propagation scenario is simulated to illustrate the applications and justify the proposed framework.
item: Thesis-Full-text
Extending Self-con guration Algorithms of Energy Constraint Wireless Sensor Networks to Emergency Environment
Gunathillake, WADA; Samarasinghe, ATLK
Wireless Sensor Networks have recently gained interest in building monitoring applications as a low cost and easy to install alternative. Some examples are smart/green buildings and emergency/rescue operations. These types of networks require that a large number of sensors be positioned easily and that they con gure themselves to perform the tasks needed without human intervention. This raises the issue of self-organization of sensor nodes. In the recent past, many researchers have investigated this topic. However, there is a lack of suitable self-organization algorithms which can be used in emer- gency monitoring applications in an indoor environment. This thesis proposes a self-organization algorithm for Wireless Sensor Networks suitable for an emer- gency detection and monitoring application by considering emergency environ- ment issues. A distributed unequal clustering algorithm with a suitable node dying pat- tern for an emergency monitoring application is proposed and simulated. The proposed algorithm optimizes the energy usage of the network and prolongs the network lifetime by multi-hop communication. The simulation result shows that the proposed algorithms prolong the network lifetime while maintaining the cov- erage of the building with existing nodes in a 2D environment. EDCR-LGRUC algorithm prolong the lifetime of the network by 1000 rounds more than the EDCR algorithm. Additional, SCAE algorithm delayed and reduced the CH fail- ures compared to EDCR. Also, the communication failure occurred due to the CH failure is reduced by 10% compared to EDCR. Moreover, 500 data rounds are optimized in the proposed multi-hop algorithm compared to EDCR-MH al- gorithm. From the application point of view, the proposed algorithm is simulated in a 3D environment. The result shows that, it achieves the same outcome as in 2D environments and that the algorithm is suitable for a wireless sensor network deployed in a multi-story building.
item: Thesis-Abstract
Geometrically constrained object tracking in non-overlapping calibrated cameras within a bayesian framework
(2016-01-16) Jayamanne, DJ; Rodrigo, R
When establishing correspondence between objects across non-overlapping cameras, the existing methods combine separate likelihoods of appearance and kinematic features in a Bayesian framework, constructing a joint likelihood to compute the probability of re-detection. So far, no method has assumed dependence between appearance and kinematic features. In this work we introduce a novel methodology to condition the location of an object on its appearance and time, without assuming independence between appearance and kinematic features, in contrast to existing work. We characterize the linear movement of objects in the unobserved region with an additive Gaussian noise model. Assuming that the cameras are affine, we transform the noise model onto the image plane of subsequent cameras. This noise model acts as a prior to improving re-detection. We have tested our hypothesis with toy car experiments and real-world camera setups. The prior constrains the search space in a subsequent camera, greatly improving the computational efficiency. Our method also has the potential to distinguish between similar-type objects, and recover correct labels when they move across cameras.
item: Thesis-Full-text
Sensor network-based indoor localization and tracking for emergency situations
(2015-09-04) Dias, SAD
Wireless Sensor Networks (WSNs) are application-specific systems, each having its own requirements related to the design. Using WSNs for emergency rescue operations is one such special application having localization of sensor nodes in a simple manner, tracking of moving nodes, usually worn by rescue workers and navigation support for rescue workers, as its major requirements. The overall objective of this research is to develop a suit of algorithms for localization, tracking and navigation of wireless sensor nodes in multistory indoor environments in emergency situations. We base our research on the DV-Hop (Distance Vector) algorithm, which is an attractive option for the localization of nodes in a wireless sensor network due to its simplicity. We carry out a comprehensive study of the DV-Hop algorithm and its variations through literature review and computer simulations. We then evaluate its performance in emergency situations, where nodes may perish, new nodes may be introduced, and communications links may be disrupted and new links set up. We then propose a new algorithm for the improvement of localization accuracy of the DV-Hop algorithm. The new algorithm is based on optimizing the Hop Size estimation in the original algorithm, which is its key source of error. We next present a new approach for target tracking in WSNs by combining the DV-Hop algorithm with Kalman filtering. The DV-Hop algorithm is used for prelocalization of the target and measurement conversion. Finally, we present a novel navigation support algorithm for rescue personnel in emergency situations by emulating virtually through WSN nodes, the lifeline used by the fire fighters. The key contribution of this work is the development of WSN localization and tracking techniques which are distributed in nature and resilient in emergency situations.
item: Thesis-Abstract
A Framework for interoperable location based services application development on mobile devices
(2015-03-01) Fernando, TMT; Dias, D
A key limitation in the present generation of Location Based Services (LBS) applications is the lack of interoperability in terms of the data sources they utilize and the devices they run on. As most LBS applications are commonly bound to a specific high-end mobile platform, they fail to run on mobile phones which use other platforms. Even though two thirds of the world population use feature phones, the majority of available LBS applications cater only to smart phones which are used by the remaining one third. Many LBS applications are also bound to a single map data provider and they have limited facilities to use already available geographic data in local contexts or at individual level. In order to address the above interoperability issues, a framework which can be utilized to generate LBS applications interoperable in terms of the devices they run and the data sources they use is proposed in this research. The interoperability related to map data is achieved by using map servers compliant with Open Geographic Consortium’s (OGC) Web Map Service (WMS) and Web Feature Service (WFS) based map servers via a Cascading Map Service. The interoperability related to devices is achieved by using a mobile web based approach. The key contribution of the research presented is a mechanism to generate LBS applications which are interoperable in these two contexts. The proposed framework is implemented using web technologies. Four sample LBS applications which make use of data from six map servers implemented using two different technologies are generated as a proof of the concept. Six different mobile phones which range from feature phones to smart phones are used for testing. Results indicate that the interoperability related to map data and devices can be achieved using the proposed approach. The framework can be used to generate interoperable LBS applications targeting the general public as well as for specific domains, for example management of resources of public utilities.
item: Thesis-Abstract
Probabilistic estimation of lanes using vehicle GPS trajectories
(2015-02-22) Manel, ERIAC; Dias, SAD; Perera, AS
This dissertation presents a novel process for generating lane-level information for road maps using a collection of trajectories from vehicles travelling on the road. The information thus generated will aid a range of Intelligent Transportation System (ITS) applications. Recently, there has been a surge of interest in research in the arena of Intelligent Transportation Systems (ITS). These systems are expected to ensure driver/passenger safety, assist the driver, support green concepts and to improve the overall efficiency and the performance of transportation systems. However, ITS require more information than what current road maps provide. Generating and refining road maps having such level of detail using the existing methods such as surveying and digitization are time consuming, costly and incompatible with the real-time and the dynamic nature of the road network. Therefore, finding new ways of generating this additional information is of high importance in making ITS a reality. The new method we propose generates lane-level information such as lane centerlines, boundaries and lane width using vehicle trajectory data. This is achieved by modeling the Probability Density Function (PDF) of trajectories across the road using the non-parametric Kernel Density Estimation (KDE). Unlike the existing methods that use Differential GPS (DGPS) data or improved GPS data, the proposed method uses ordinary GPS data obtained from vehicles moving along the road. It does not require any information regarding the road parameters and is completely automatic. Furthermore, it is completely independent of the lane/road width and does not use stringent assumptions on lane parallelism and constant lane width. In particular, it estimates the locations of lane centers, locations of lane boundaries and lane width. The proposed method for calculating the lane centers was proven to be successful in different road geometries such as straight sections, curved sections and sections with lane splits and merges. The method proposed for calculating lane boundaries produced good results when there are no gaps in between lanes. The lane width calculated using the proposed method is compatible with the recorded standard lane with of the chosen road. Keywords: GPS, Lane level maps, Kernel density estimation, Kernel bandwidth