No description available.

This study was carried out in collaboration with Tracab, a brand leader in real-time digital sports data. As a result, the application field is centred on sports analytics. The technology, for instance, consists of multiple cameras that capture a football pitch in a panoramic setup. The alignment of two or more cameras in both a spatial and temporal manner is referred to as sub-frame synchronisation. Because the cameras are already in the same geometric coordinates, only temporal synchronisation will be addressed in this project. The main method for retrieving the desynchronisation information that affects the cameras is based on optical flow. The off-sync cameras' spacial information is then synthesised to the time required by the synchronisation constraint using motion interpolation. In addition, the created system is compared to a real-time intermediate flow interpolation approach. The latter method relies on machine learning techniques, whereas this study focuses on more traditional methods. The metrics Peak Signal-to-Noise Ratio and Structural Similarity Index Measure are used to address the quality criteria required by this subject of study. Furthermore, visually perceived quality is examined to identify differences between measured and perceived quality. The results reveal that in every realistic situation investigated, temporal synchronisation can be addressed by an error measure of less than 1ms. The frame synthesis stage, on the other hand, fails to accurately estimate complicated scenarios, while the machine learning approach stands out. The implemented approach, on the other hand, addresses fast-moving objects with greater precision. Furthermore, the machine learning approach is unable to interpolate intermediate frames in arbitrary time steps, which is critical for the project's application. Finally, considering the lack of real-time computational speed and the quality achieved by machine learning approaches, more research is required in these directions. / Denna studie genomfördes i samarbete med Tracab, en marknadsledare inom digital sportdata levererad i realtid. Studiens applikationsområde kommer där av centreras kring sportdata där två eller flera kameror filmar en fotbollsplan i ett videopanorama. Kamerasynkroniseringen måste ske både spatialt och temporalt. Eftersom kamerorna har samma position kommer endast den temporala synkronisering tas upp i detta projekt. Den övergripande metoden för att göra detta är baserat på optiskt flöde. Data från en ej synkroniserad kamera syntetiseras via en synkroniseringkonstant mha. rörelseinterpolering. Detta jämförs även mot ett tillvägagångssätt som bygger på maskininlärning medan man i denna studie fokuserar på en mer traditionell lösningsmetod. Mätvärdena Peak Signal-to-Noise Ratio och Structural Similarity Index Measure används som kvalitetskriteria. Även visuellt upplevd kvalitet undersöks för att identifiera skillnaden mellan mätt och upplevd kvalitet. Resultatet visar att vid realistiska situationer kan den temporala synkroniseringen beräknas till under 1ms. Den syntetiserade datan lyckas dock inte estimera komplicerade situationer, medan maskininlärningsmetoden presterar bra. Dock så klarar studiens lösningsmetod att bättre generera objekt i snabb rörelse. Vidare så kan inte maskininlärningsmetoden generera video med en godtycklig tidförskjutning, något som är avgörande för projektets tillämpningsområde. Slutligen, med tanke på svårigheter i realtidsberäkning kontra kvaliteten hos maskin- inlärningsmetoder krävs därför mer forskning inom området.

La profondeur possible d’imagerie en laser-scanning microscopie est limitée non seulement par la distance de travail des lentilles de objectifs mais également par la dégradation de l’image causée par une atténuation et une diffraction de la lumière passant à travers l’échantillon. Afin d’étendre cette limite, il est possible, soit de retourner le spécimen pour enregistrer les images depuis chaque côté, or couper progressivement la partie supérieure de l’échantillon au fur et à mesure de l‘acquisition. Les différentes images prises de l’une de ces manières doivent ensuite être combinées pour générer un volume unique. Cependant, des mouvements de l’échantillon durant les procédures d’acquisition engendrent un décalage non seulement sur en translation selon les axes x, y et z mais également en rotation autour de ces même axes, rendant la fusion entres ces multiples images difficile. Nous avons développé une nouvelle approche appelée 2D-SIFT-in-3D-Space utilisant les SIFT (scale Invariant Feature Transform) pour atteindre un recalage robuste en trois dimensions de deux images. Notre méthode recale les images en corrigeant séparément les translations et rotations sur les trois axes grâce à l’extraction et l’association de caractéristiques stables de leurs coupes transversales bidimensionnelles. Pour évaluer la qualité du recalage, nous avons également développé un simulateur d’images de laser-scanning microscopie qui génère une paire d’images 3D virtuelle dans laquelle le niveau de bruit et les angles de rotations entre les angles de rotation sont contrôlés avec des paramètres connus. Pour une concaténation précise et naturelle de deux images, nous avons également développé un module permettant une compensation progressive de la luminosité et du contraste en fonction de la distance à la surface de l’échantillon. Ces outils ont été utilisés avec succès pour l’obtention d’images tridimensionnelles de haute résolution du cerveau de la mouche Drosophila melanogaster, particulièrement des neurones dopaminergiques, octopaminergiques et de leurs synapses. Ces neurones monoamines sont particulièrement important pour le fonctionnement du cerveau et une étude de leur réseau et connectivité est nécessaire pour comprendre leurs interactions. Si une évolution de leur connectivité au cours du temps n’a pas pu être démontrée via l’analyse de la répartition des sites synaptiques, l’étude suggère cependant que l’inactivation de l’un de ces types de neurones entraine des changements drastiques dans le réseau neuronal. / Although laser scanning microscopy is a powerful tool for obtaining thin optical sections, the possible depth of imaging is limited by the working distance of the microscope objective but also by the image degradation caused by the attenuation of both excitation laser beam and the light emitted from the fluorescence-labeled objects. Several workaround techniques have been employed to overcome this problem, such as recording the images from both sides of the sample, or by progressively cutting off the sample surface. The different views must then be combined in a unique volume. However, a straightforward concatenation is often not possible, because the small rotations that occur during the acquisition procedure, not only in translation along x, y and z axes but also in rotation around those axis, making the fusion uneasy. To address this problem we implemented a new algorithm called 2D-SIFT-in-3D-Space using SIFT (scale Invariant Feature Transform) to achieve a robust registration of big image stacks. Our method register the images fixing separately rotations and translations around the three axes using the extraction and matching of stable features in 2D cross-sections. In order to evaluate the registration quality, we created a simulator that generates artificial images that mimic laser scanning image stacks to make a mock pair of image stacks one of which is made from the same stack with the other but is rotated arbitrarily with known angles and filtered with a known noise. For a precise and natural-looking concatenation of the two images, we also developed a module progressively correcting the sample brightness and contrast depending on the sample surface. Those tools we successfully used to generate tridimensional high resolution images of the fly Drosophila melanogaster brain, in particular, its octopaminergic and dopaminergic neurons and their synapses. Those monoamine neurons appear to be determinant in the correct operating of the central nervous system and a precise and systematic analysis of their evolution and interaction is necessary to understand its mechanisms. If an evolution over time could not be highlighted through the pre-synaptic sites analysis, our study suggests however that the inactivation of one of these neuron types triggers drastic changes in the neural network.

Panoramas or stitched images are used in topographical mapping, panoramic 3D reconstruction, deep space exploration image processing, medical image processing, multimedia broadcasting, system automation, photography and other numerous fields. Generating real-time panoramic images in small embedded computer is of particular importance being lighter, smaller and mobile imaging system. Moreover, this type of lightweight panoramic imaging system is used for different types of industrial or home inspection. A real-time handheld panorama imaging system is developed using embedded real-time Linux as software module and Gumstix Overo and PandaBoard ES as hardware module. The proposed algorithm takes 62.6602 milliseconds to generate a panorama frame from three images using a homography matrix. Hence, the proposed algorithm is capable of generating panorama video with 15.95909365 frames per second. However, the algorithm is capable to be much speedier with more optimal homography matrix. During the development, Ångström Linux and Ubuntu Linux are used as the operating system with Gumstix Overo and PandaBoard ES respectively. The real-time kernel patch is used to configure the non-real-time Linux distribution for real-time operation. The serial communication software tools C-Kermit, Minicom are used for terminal emulation between development computer and small embedded computer. The software framework of the system consist UVC driver, V4L/V4L2 API, OpenCV API, FFMPEG API, GStreamer, x264, Cmake, Make software packages. The software framework of the system also consist stitching algorithm that has been adopted from available stitching methods with necessary modification. Our proposed stitching process automatically finds out motion model of the Spherical camera system and saves the matrix in a look file. The extracted homography matrix is then read from look file and used to generate real-time panorama image. The developed system generates real-time 180° view panorama image from a spherical camera system. Beside, a test environment is also developed to experiment calibration and real-time stitching with different image parameters. It is able to take images with different resolutions as input and produce high quality real-time panorama image. The QT framework is used to develop a multifunctional standalone software that has functions for displaying real-time process algorithm performance in real-time through data visualization, camera system calibration and other stitching options. The software runs both in Linux and Windows. Moreover, the system has been also realized as a prototype to develop a chimney inspection system for a local company. / Main Uddin-Al-Hasan, E-mail: main.hasan@gmail.com

This Master's Thesis delves into the application of topological regularization techniques and relative latent representations within the realm of zero-shot model stitching. Building upon the prior work of Moschella et al. (2022) that introduces relative latent representations to enhance the similarities between latent spaces of different models, we incorporate the approach of Hofer et al. (2021), which combines Topological Data Analysis (TDA) and Machine Learning techniques for topological densification of class distributions in the latent space. The main research objective is to investigate the impact of topological regularization on zero-shot stitching performance when employing relative latent representations. Theoretical foundations for the relative transformation are established based on the intertwiner groups of activation functions. Empirical analyses are conducted to validate the assumptions underlying the construction of the relative transformation in the latent space. Moreover, experiments are performed on a Large Language Model trained on multilingual Amazon Reviews datasets to evaluate the effectiveness of zero-shot stitching while using the topological densification technique and the relative transformation. The findings indicate that the proposed methodologies can enhance the performance of multilingual model stitching. Specifically, enforcing the relative transformation to preserve the H0 homology death times distributions proves beneficial. Additionally, the presence of similar topological features plays a crucial role in achieving higher model compatibility. However, a more in-depth exploration of the geometric properties of the post-relative transformation latent space is necessary to further improve the topological densification technique. Overall, this work contributes to the emerging field of Topological Machine Learning and provides valuable insights for researchers in transfer learning and representation learning domains. / Denna masteruppsats undersöker tillämpningen av topologiska regleringstekniker och relativa latenta representationer inom området för zero-shot model stitching. Genom att bygga vidare på tidigare arbete av Moschella et al. (2022), som introducerade relativa latenta representationer för att förbättra likheterna mellan latenta rummet hos olika modeller, inkorporerar vi tillvägagångssättet av Hofer et al. (2021), som kombinerar topologisk dataanalys (TDA) och maskininlärningstekniker för topologisk ``förtätning'' av klassfördelningar i det latenta utrymmet. Den huvudsakliga forskningsuppgiften är att undersöka effekten av topologisk reglering på zero-shot model stitching-prestanda när man använder relativa latenta representationer. Teoretiska grunder för den relativa transformationen etableras baserat på intertwinergrupperna för aktiveringsfunktioner. Empiriska analyser genomförs för att validera antagandena som ligger till grund för konstruktionen av den relativa transformationen i det latenta rummen. Dessutom utförs experiment på en stor språkmodell tränad på multilinguella Amazon Reviews-dataset för att utvärdera effektiviteten hos zero-shot model stitching med Hofer's topologiska reglering och relativa transformation. Resultaten visar att de föreslagna metoderna kan förbättra prestationen hos zero-shot model stitching för flerspråkiga modeller. Specifikt är det fördelaktigt att tvinga den relativa transformationen att bevara H0 homologins dödstidsfördelningar. Dessutom spelar närvaron av liknande topologiska egenskaper en avgörande roll för att uppnå högre modellkompatibilitet. Dock krävs en mer ingående utforskning av de geometriska egenskaperna hos det latenta utrymmet efter den relativa transformationen för att ytterligare förbättra Hofer's topologiska reglering. Sammanfattningsvis bidrar detta arbete till det framväxande området Topologisk Maskininlärning och ger värdefulla insikter för forskare inom ``transfer-inlärning'' och representationsinlärningsdomäner.

This document deals with an image points of interest detection possibilities, especially corner detectors. Many applications which are interested in computer vision needs these points as their necessary step in the image processing. It describes the reasons why it is so useful to find these points and shows some basic methods to find them. There are compared features of these methods at the end.