Lund University Publications
Not a member yet
244244 research outputs found
Sort by
Prototype of Image Preprocessing and Analysis System Supporting UAV Camera Guidance
No full textThis work presents the architecture of a prototype software system for image preprocessing and analysis designed to support the guidance of an auxiliary UAV camera. The study begins with an overview of typical UAV onboard elec-tronics, highlighting the electro-optical equipment, which includes a navigation camera and an auxiliary camera with a narrower field of view. Automatic image analysis from the navigation camera enables the detection of potential areas of in-terest and directs the auxiliary camera accordingly. The magnified image from the auxiliary camera facilitates decision-making and enhances the performance of the UAV pilot-operator during extended missions.Various object detection models are reviewed, and a single-stage YOLO ar-chitecture is selected as the most suitable for the task. Building on previous re-sults, where physical alignment of the auxiliary camera with the center of the nav-igation camera’s frame was implemented in MATLAB/Simulink, additional sys-tem components have been developed in the same environment. These compo-nents perform image preprocessing, object localization and classification, and de-termine the target zone number for positioning the auxiliary camera.The prototype system, developed in a mixed software-hardware environment, allows rapid modification and testing of individual modules. Furthermore, using MATLAB’s built-in code generation and deployment tools, the software can be adapted for UAV onboard computers using high-level programming languages such as C++ or Python
Unveiling impulse buying patterns in travel live-streaming through the lens of social cognitive theory
No full textGrounded in the social cognitive theory, the main aim of this study is to explore the impact of telepresence, social presence, and emotional engagement on impulsive buying tendencies within the realm of travel live-streaming (TLS). To examine further, the moderating effect of guidance shopping was included. An online questionnaire was distributed to 332 live-streaming viewers from China, and the collected data were analyzed using partial least squares structural equation modeling (PLS-SEM). The findings indicate that telepresence and social presence exert significant influences on emotional engagement and impulsive buying tendencies. Notably, the study identifies the mediating effect of emotional engagement and the moderating effect of guidance shopping. In practice, this research highlights the importance of streamers leveraging TLS platform functionalities to amplify the sense of “presence” and cultivate stronger emotional connections with viewers, thereby stimulating impulse buying. Additionally, streamers are encouraged to guide viewers in their decision-making process, addressing concerns to elevate engagement levels
Starch microsphere preparation and phase behaviour in aqueous two-phase systems – effect of continuous-phase polymer
No full textWhen producing starch microspheres in aqueous two-phase systems (ATPS), a continuous-phase polymer is employed to induce segregative phase separation. This creates a continuous polymer-rich phase, in which starch-rich phase droplets can be dispersed and crystallised into solid, semi-crystalline starch microspheres. This study aims to explore how polymers with different chemical structures and sizes affect phase behaviour in starch-based ATPS and the formation of starch microspheres. The polymers studied were polyethylene glycol (PEG), polyvinylpyrrolidone (PVP), poly(2-ethyl-2-oxazoline) (PEtOx) and hydroxypropyl methylcellulose (HPMC). To reach a starch concentration in the starch-rich phase (55 %) that yielded solid starch microspheres, less PEG (32 %) than PVP (39 %) or PEtOx (42 %) were required in the continuous phase. The sizes of the polymers had little or no effect on the water distribution in the ATPS. The governing factor is the water activity within the system. The water activity in the starch-rich phase must be low enough to allow for crystallisation within the predetermined time frame. Accordingly, the polymer concentration must be selected so that this water activity condition is reached. The viscosity of the continuous phase significantly impacted the microsphere production. It needs to be high enough to prevent the coalescence of starch droplets as they transition to solid gels, but not so high that it hinders emulsification. While polymer choice affected particle size and aggregation, it did not notably change the interior structure, surface morphology, crystal type, or thermal behaviour of the dried starch microspheres. This opens up the possibility to optimise the production of microspheres by changing polymers, while tuning the starch crystallisation through control of the water activity in the starch phase
Oxy-ethane combustion : Measurements of the laminar burning velocities and kinetic insights into CO2 effects
No full textOxy-fuel combustion is regarded as an underpinning technology for carbon capture and utilization. Ethane (C2H6) is the primary non-methane component in natural gas and a fundamental C2 fuel. In this work, laminar burning velocities (LBV) of C2H6/O2/CO2 mixtures were measured using the heat flux method at atmospheric pressure, spanning initial temperatures of 308–348 K, equivalence ratios of 0.6–1.5, and CO2 fractions of 0.16–0.75. The temperature dependence of the LBV of C2H6/O2/CO2 mixtures was first determined, and the consistency of the new measurements was validated. Evaluations of three kinetic models (Oxymech 2.0, Konnov 2023, Alzueta 2015) demonstrated that Oxymech 2.0 accurately predicts LBVs and ignition delay times (IDTs) across all conditions. Nonlinear variation of LBV with CO2 amount in the mixture was observed from both experiments and simulations. Decoupling thermal, transport, and chemical effects revealed that the thermal and transport inhibiting effects of CO2 linearly increase with CO2 concentration, while CO2′s chemical inhibition initially strengthens (CO2 range: 0.16–0.39), then plateaus (CO2: 0.39–0.65). This nonlinearity stems from the reaction CO + OH=H + CO2 depleting H radicals, suppressing chain-branching, while the reduced ratio of 2CH3=H + C2H5 versus C2H4 + H(+M)=C2H5(+M) (promoting flame propagation) offsets inhibition at higher CO2 fractions
Pore-scale numerical investigation of water–gas flow and heat transport in gas diffusion layers with varying fiber/additive content and hydrophobicity
No full textAdditives such as binders and hydrophobic agents are commonly introduced into the fibrous Gas Diffusion Layer (GDL) of proton exchange membrane fuel cells to enhance mechanical strength and facilitate water management. However, the effect of additive/fiber content and surface wettability on water removal, oxygen diffusion, and heat conduction remains insufficiently understood. In this work, we develop a stochastic GDL reconstruction framework with systematically varied fiber and additive content. The reconstructed structures are analyzed through pore–throat network extraction, interface-resolved two-phase flow simulations, as well as oxygen diffusion and heat conduction simulations under dry and partially saturated conditions. The variation in surface wettability caused by the coating of hydrophobic additives is simulated by the contact angle. The results reveal that increased fiber content significantly restricts pore space, thereby weakening oxygen diffusivity and increasing breakthrough pressure, while having a limited impact on stabilized water saturation and thermal conductivity. Additives, particularly at high loadings, reduce pore connectivity and gas transport, though enhanced hydrophobicity partially mitigates these effects. Oxygen diffusivity is found to be particularly sensitive to changes in effective pore space caused by additive inclusion and water occupation. These findings present a comprehensive quantitative perspective on how additive design modulates GDL transport properties and provide a simulation-based framework for optimizing fuel cell GDL microstructure
Does surgical history matter? A register-based study of 94 000 individuals from the Swedish osteoarthritis register on clinical profiles and outcomes of first-line treatment for knee osteoarthritis
No full textBackground There is limited knowledge on whether prior knee surgery impacts the clinical profile and treatment outcomes for individuals with knee osteoarthritis (OA). Objectives The study aimed to (i) compare individual and clinical characteristics, and (ii) evaluate whether outcomes of first-line treatment differ between individuals with prior knee surgery and those without. Methods This study used data from the Swedish Osteoarthritis Register to compare individuals with and without prior knee surgery. Independent t-tests and chi-square tests analyzed characteristics, while linear and logistic regressions assessed group differences at the 3-month follow-up. Results Of the 94 116 individuals included, 15 637 (17 %) had prior knee surgery. At baseline, those with prior knee surgery were more likely to be male (48 % vs. 29 %), younger (mean age 63 vs. 67 years), and meet physical activity recommendations (69 % vs. 66 %). At the 3-month follow-up, individuals with prior knee surgery had higher odds of expressing willingness to undergo surgery (OR 1.50 [95 % CI, 1.40 to 1.60]), experiencing walking difficulties (OR 1.24 [95 % CI, 1.19 to 1.30]), and to reaching the recommended level of physical activity (OR 1.21 [95 % CI, 1.14 to 1.27]). Conclusion The findings suggest that individuals with knee OA and a history of prior knee surgery may represent a specific clinical phenotype, characterized by a younger age, male sex, and higher levels of physical activity when entering first-line treatment. Despite positive treatment responses in both groups, individuals with a history of surgery more often reported ongoing clinical features at the 3-month follow-up
Attacking Single-Cycle Ciphers on Modern FPGAs : Featuring Explainable Deep Learning
No full textIn this paper, we revisit the question of key recovery using side-channel analysis for unrolled, single-cycle block ciphers. In particular, we study the Princev2 cipher. While it has been shown vulnerable in multiple previous studies, those studies were performed on side-channel friendly ASICs or older FPGAs (e.g., Xilinx Virtex II on the SASEBO-G board), and using mostly expensive equipment. We start with the goal of exploiting a cheap modern FPGA and board using power traces from a cheap oscilloscope. Particularly, we use Xilinx Artix 7 on the Chipwhisperer CW305 board and PicoScope 5000A, respectively. We split our study into three parts. First, we show that the new set-up still exhibits easily detectable leakage, using a non-specific t-test. Second, we replicate attacks from older FPGAs. Namely, we start with the attack by Yli-Mäyry et al., which is a simple chosen plaintext correlation power analysis attack using divide and conquer. However, we demonstrate that even this simple, powerful attack does not work, demonstrating a peculiar behavior. We study this behavior using a stochastic attack that attempts to extract the leakage model, and we show that models over a small part of the state are inconsistent and depend on more key bits than what is expected. We also attempt classical template attacks and get similar results. To further exploit the leakage, we employ deep learning techniques and succeed in key recovery, albeit using a large number of traces. We perform the explainability technique called Key Guessing Occlusion (KGO) to detect which points the neural networks exploit. When we use these points as features for the classical template attack, although it did not recover the secret key, its performance improves compared to other feature selection techniques
Guiding Light to Detect Life: Nanowires for Optical Biosensing
No full textSemiconductor nanowires provide a versatile platform for fluorescence-based detection enabled by their high refractive index and waveguiding properties, which amplify excitation and emission from surface-bound fluorophores. Together with their large surface-to-volume ratio, these features enable enhanced and even single-molecule sensitivity.This thesis explores the use of gallium phosphide (GaP) and silicon (Si) nanowires through integrated experimental and computational studies, demonstrating their capabilities for sensitive and quantitative biosensing applications. Using single-emitter localization on well-spaced vertical nanowires combined with brightfield microscopy and systematic analysis, detection ranges can extend over five orders of magnitude, reaching femtomolar levels in streptavidin-biotin assays. Similarly, immobilizing fluorescent molecular beacons on nanowires improves signal-to-background contrast and allows direct oligonucleotide detection at sub-nanomolar concentrations. The nanowire geometry also facilitates extraction of off-plane molecular positions: by modeling point spread functions and training convolutional neural networks on simulated datasets, axial localization accuracies below 100 nm are achieved, enabling tracking of labeled DNA diffusing in supported lipid bilayers.Nanowires also function effectively in complex sample environments. When embedded in polymer matrices and imaged through transparent substrates, they overcome scattering and absorption in opaque media, allowing sub-nanomolar detection of fluorescently labeled proteins in whole blood, lipid emulsions, and powdered milk without sample processing. Overall, this thesis expands the application scope of semiconductor nanowires for optical biosensing, demonstrating single-molecule detection, extended dynamic range, direct oligonucleotide sensing, three-dimensional localization, and operation in opaque biological samples