Cette thèse vise à améliorer la résolution en trois dimensions grâce à une technique récente d’imagerie : la microscopie tomographique diffractive (MTD). Son principe est d’éclairer l’objet successivement sous différents angles en lumière cohérente, de détecter le champ diffracté en phase et en amplitude, et de reconstruire la carte 3D de permittivité de l’objet par un algorithme d’inversion. La MTD s’est avérée capable de combiner plusieurs modalités utiles pour la microscopie sans marquage, telles que plein champ, champ sombre, à contraste de phase, confocale, ou encore la microscopie à synthèse d’ouverture 2D ou 3D. Toutes sont basées sur des approximations scalaires et linéaires, ce qui restreint leur domaine d’application pour restituer l’objet de manière quantitative. A l’aide d’une inversion numérique rigoureuse prenant en compte la polarisation du champ et le phénomène de diffusion multiple, nous sommes parvenus à reconstruire la carte 3D de permittivité d’objets avec une résolution de λ/4. Une amélioration supplémentaire la portant à λ/10 a été rendue possible par l’insertion d’information a priori sur l’objet dans l’algorithme d’inversion. Enfin, la résolution axiale est moins bonne du fait de l’asymétrie des schémas d’illumination et de détection dans les microscopes. Pour s’affranchir de cette limitation, une configuration de tomographie assistée par miroir a été implémentée et a mis en évidence un pouvoir de séparation axial meilleur que λ/2. Au final, la MTD s’est illustrée comme un outil de caractérisation puissant pour reconstruire en 3D les objets ainsi que leurs indices optiques, à des résolutions bien supérieures à celles des microscopes conventionnels. / This PhD thesis is devoted to the three-dimensional isotropic resolution improvement using optical tomographic diffraction microscopy (TDM), an emerging optical microscope technique. The principle is to illuminate the sample successively with various angles of coherent light, collect the complex (amplitude and phase) diffracted field and reconstruct the sample 3D permittivity map through an inversion algorithm. A single TDM measurement was shown to combine several popular microscopy techniques such as bright-field microscope, dark-field microscope, phase-contrast microscope, confocal microscope, 2D and 3D synthetic aperture microscopes. All rely on scalar and linear approximations that assume a linear link between the object and the field diffracted by it, which limit their applicability to retrieve the object quantitatively. Thanks to a rigorous numerical inversion of the TDM diffracted field data which takes into account the polarization of the field and the multiple scattering process, we were able to reconstruct the 3D permittivity map of the object with a λ/4 transverse resolution. A further improvement to λ/10 transverse resolution was achieved by providing a priori information about the sample to the non-linear inversion algorithm. Lastly, the poor axial resolution in microscopes is due to the fundamental asymmetry of illumination and detection. To overcome this, a mirror-assisted tomography configuration was implemented, and has demonstrated a sub-λ/2 axial resolution capability. As a result, TDM can be seen as a powerful tool to reconstruct objects in three-dimensions with their optical material properties at resolution far superior to conventional microscopes.

High quality image /video has become an integral part in our day-to-day life ranging from many areas of science, engineering and medical diagnosis. All these imaging applications call for high resolution, properly focused and crisp images. However, in real situations obtaining such a high quality image is expensive, and in some cases it is not practical. In imaging systems such as digital camera, blur and noise degrade the image quality. The recorded images look blurred, noisy and unable to resolve the finer details of the scene, which are clearly notable under zoomed conditions. The post processing techniques based on computational methods extract the hidden information and thereby improve the quality of the captured images. The study in this thesis focuses on deconvolution and eventually blind de-convolution problem of a single frame captured at low light imaging conditions arising from digital photography/surveillance imaging applications. Our intention is to restore a sharp image from its blurred and noisy observation, when the blur is completely known/unknown and such inverse problems are ill-posed/twice ill-posed. This thesis consists of two major parts. The first part addresses deconvolution/blind deconvolution problem using Bayesian approach with fuzzy logic based gradient potential as a prior functional. In comparison with analog cameras, artifacts are visible in digital cameras when the images are enlarged and there is a demand to enhance the resolution. The increased resolution can be in spatial, temporal or even in both the dimensions. Super resolution reconstruction methods reconstruct images/video containing spectral information beyond that is available in the captured low resolution images. The second part of the thesis addresses resolution enhancement of observed monochromatic/color images using multiple frames of the same scene. This reconstruction problem is formulated in Bayesian domain with an aspiration of reducing blur, noise, aliasing and increasing the spatial resolution. The image is modeled as Markov random field and a fuzzy logic filter based gradient potential is used to differentiate between edge and noisy pixels. Suitable priors are adaptively applied to obtain artifact free/reduced images. In this work, all our approaches are experimentally validated using standard test images. The Matlab based programming tools are used for carrying out the validation. The performance of the approaches are qualitatively compared with results of recently proposed methods. Our results turn out to be visually pleasing and quantitatively competitive.

This thesis deals with image colorization and image super-resolution using neural networks. It briefly explains neural networks principles and summarizes current approaches in this domain. It also describes the design, implementation and training of various neural network architectures. The best implemented architecture can colorize images, in particular, works well with outdoor areas. The architecture for image super-resolution with residual blocks that was trained with a perceptual loss function performs a double increase in image resolution (4x more pixels in total). Part of this thesis is also an implementation of a web application that uses trained models for image modification.

This diploma thesis deals with image super-resolution with conservation of good quality. Firstly, there are described state of the art methods dealing with this problem, as well as principles of neural networks with focus on convolutional ones. Finally, there is described a few models of convolutional neural network for image super-resolution to double size, which have been trained, tested and compared on newly created database with pictures of people.

VINTERA, J. Spectral anlysis with superesolution. Brno: University of Technology, The Faculty of Electrical Engineering and Communication, 2008. 85 p. Master’s thesis. This thesis deals with the topic of super-resolution spectral analysis in the Signal Processing Toolset. The Signal Processing Toolset is a software component of the LabVIEW 8.1. program equipment. The thesis consists of three main parts. In the first part the basic theoretic concepts of the Model-Based Frequency Analysis are described. The second part serves as a user manual for the super-resolution spectral analysis in the Signal Processing Toolset. The last part describes the application of the theory introduced in the first part, by means of testing the properties of the methods used by the Toolset.

Computer vision is a key component of any autonomous system. Real world computer vision applications rely on a proper and accurate detection and classification of objects. A detection algorithm that doesn’t guarantee reasonable detection accuracy is not applicable in real time scenarios where safety is the main objective. Factors that impact detection accuracy are illumination conditions and image resolution. Both contribute to degradation of objects and lead to low classifications and detection accuracy. Recent development of Convolutional Neural Networks (CNNs) based algorithms offers possibilities for low-light (LL) image enhancement and super resolution (SR) image generation which makes it possible to combine such models in order to improve image quality and increase detection accuracy. This thesis evaluates different CNNs models for SR generation and LL enhancement by comparing generated images against ground truth images. To quantify the impact of the respective model on detection accuracy, a detection procedure was evaluated on generated images. Experimental results evaluated on images selected from NoghtOwls and Caltech Pedestrian datasets proved that super resolution image generation and low-light image enhancement improve detection accuracy by a substantial margin. Additionally, it has been proven that a cascade of SR generation and LL enhancement further boosts detection accuracy. However, the main drawback of such cascades is related to an increased computational time which limits possibilities for a range of real time applications. / Datorseende är en nyckelkomponent i alla autonoma system. Applikationer för datorseende i realtid är beroende av en korrekt detektering och klassificering av objekt. En detekteringsalgoritm som inte kan garantera rimlig noggrannhet är inte tillämpningsbar i realtidsscenarier, där huvudmålet är säkerhet. Faktorer som påverkar detekteringsnoggrannheten är belysningförhållanden och bildupplösning. Dessa bidrar till degradering av objekt och leder till låg klassificerings- och detekteringsnoggrannhet. Senaste utvecklingar av Convolutional Neural Networks (CNNs) -baserade algoritmer erbjuder möjligheter för förbättring av bilder med dålig belysning och bildgenerering med superupplösning vilket gör det möjligt att kombinera sådana modeller för att förbättra bildkvaliteten och öka detekteringsnoggrannheten. I denna uppsats utvärderas olika CNN-modeller för superupplösning och förbättring av bilder med dålig belysning genom att jämföra genererade bilder med det faktiska data. För att kvantifiera inverkan av respektive modell på detektionsnoggrannhet utvärderades en detekteringsprocedur på genererade bilder. Experimentella resultat utvärderades på bilder utvalda från NoghtOwls och Caltech datauppsättningar för fotgängare och visade att bildgenerering med superupplösning och bildförbättring i svagt ljus förbättrar noggrannheten med en betydande marginal. Dessutom har det bevisats att en kaskad av superupplösning-generering och förbättring av bilder med dålig belysning ytterligare ökar noggrannheten. Den största nackdelen med sådana kaskader är relaterad till en ökad beräkningstid som begränsar möjligheterna för en rad realtidsapplikationer.

<div>The research includes machine Learning and Deep Learning Approaches to Print Defect Detection, Face Set Recognition and Face Alignment, and Visual-Enhancement in Space and Time. This thesis consists of six parts which are related to 6 projects:</div><div><br></div><div>In Chapter 1, the first project focuses on detection of local printing defects including gray spots and solid spots. We propose a coarse-to-fine method to detect local defects in a block-wise manner and aggregate the blockwise attributes to generate the feature vector of the whole test page for a further ranking task. In the detection part, we first select candidate regions by thresholding a single feature. Then more detailed features of candidate blocks are calculated and sent to a decision tree that is previously trained on our training dataset. The final result is given by the decision tree model to control the false alarm rate while maintaining the required miss rate.</div><div><br></div><div>Chapter 2 introduces face set recognition and Chapter 3 is about face alignment. In order to reduce the computational complexity of comparing face sets, we propose a deep neural network that can compute and aggregate the face feature vectors with different weights. As for face alignment, our goal is to solve the jittering of landmark locations when applied on video. We propose metrics and corresponding methods around this goal.</div><div><br></div><div>In recent years, mobile photography has become increasingly prevalent in our lives with social media due to its high portability and convenience. However, many challenges still exist in distributing high-quality mobile images and videos under the limit of data capacity, hardware storage, and network bandwidth. Therefore, we have been exploring enhancement techniques to improve the image and video qualities, considering both effectiveness and efficiency for a wide variety of applications, including WhatsApp, Portal, TikTok, even the printing industry. Chapter 4 introduces single image super-resolution to handle real-world images with various degradations, and its influence on several downstream high-level computer vision tasks. Next, Chapter 5 studies on headshot image restoration with multiple references, which is an application of visual enhancement under more specific scenarios. Finally, as a step towards the temporal domain enhancement, the Zooming SlowMo framework for fast and accurate space-time video super-resolution will be introduced in Chapter 6.</div>

10 SUMMARY Proteins represent one of the basic building blocks of all organisms. To understand their function at the molecular level is one the critical goals of current biological, biochemical and biophysical research. It is important to characterise all aspects that affect the localisation of proteins into different compartments with specific functions, the dynamic structure of proteins and their role in multiprotein assemblies, because altering these properties can lead to various diseases. Most of the proteomic studies are nowadays performed using biochemical approaches that allow us to study multicellular organism or tissue at once. The disadvantage of these methods is complex preparation of sample and the need for a large number of cells, which leads to the loss of information at the molecular level and in individual cells. On the contrary, microscopy can provide rather detailed information about proteins of interest and at the level of a single cell. A variety of fluorescence microscopy methods in combination with recombinant DNA techniques were applied to elucidate subcellular localisation of transmembrane adaptor proteins (TRAPs) in human lymphocytes and their nanoscopic organisation at the plasma membrane. Linker of activation of T lymphocytes (LAT), phosphoprotein associated with...

Generative Adversarial Networks have been used to obtain state-of-the-art results for low-level computer vision tasks like single image super-resolution, however, they are notoriously difficult to train due to the instability related to the competing minimax framework. Additionally, traditional ensembling mechanisms cannot be effectively applied with these types of networks due to the resources they require at inference time and the complexity of their architectures. In this thesis an alternative method to create ensembles of individual, more stable and easier to train, models by using interpolations in the parameter space of the models is found to produce better results than those of the initial individual models when evaluated using perceptual metrics as a proxy of human judges. This method can be used as a framework to train GANs with competitive perceptual results in comparison to state-of-the-art alternatives. / Generative Adversarial Networks (GANs) har använts för att uppnå state-of-the- art resultat för grundläggande bildanalys uppgifter, som generering av högupplösta bilder från bilder med låg upplösning, men de är notoriskt svåra att träna på grund av instabiliteten relaterad till det konkurrerande minimax-ramverket. Dessutom kan traditionella mekanismer för att generera ensembler inte tillämpas effektivt med dessa typer av nätverk på grund av de resurser de behöver vid inferenstid och deras arkitekturs komplexitet. I det här projektet har en alternativ metod för att samla enskilda, mer stabila och modeller som är lättare att träna genom interpolation i parameterrymden visat sig ge bättre perceptuella resultat än de ursprungliga enskilda modellerna och denna metod kan användas som ett ramverk för att träna GAN med konkurrenskraftig perceptuell prestanda jämfört med toppmodern teknik.

The riveting interplay of industrialization, informalization, and exponential technological growth of recent years has shifted the attention from classical machine learning techniques to more sophisticated deep learning approaches; yet its intrinsic black-box nature has been impeding its widespread adoption in transparency-critical operations. In this rapidly evolving landscape, where the symbiotic relationship between research and practical applications has never been more interwoven, the contribution of this paper is twofold: advancing gradient faithfulness of CAM methods and exploring new frontiers beyond it. In the first part, we theorize three novel gradient-based CAM formulations, aimed at replacing and superseding traditional Grad-CAM-based methods by tackling and addressing the intricately and persistent vanishing and saturating gradient problems. As a consequence, our work introduces novel enhancements to Grad-CAM that reshape the conventional gradient computation by incorporating a customized and adapted technique inspired by the well-established and provably Expected Gradients’ difference-from-reference approach. Our proposed techniques– Expected Grad-CAM, Expected Grad-CAM++and Guided Expected Grad-CAM– as they operate directly on the gradient computation, rather than the recombination of the weighing factors, are designed as a direct and seamless replacement for Grad-CAM and any posterior work built upon it. In the second part, we build on our prior proposition and devise a novel CAM method that produces both high-resolution and class-discriminative explanation without fusing other methods, while addressing the issues of both gradient and CAM methods altogether. Our last and most advanced proposition, Hyper Expected Grad-CAM, challenges the current state and formulation of visual explanation and faithfulness and produces a new type of hybrid saliencies that satisfy the notion of natural encoding and perceived resolution. By rethinking faithfulness and resolution is possible to generate saliencies which are more detailed, localized, and less noisy, but most importantly that are composed of only concepts that are encoded by the layerwise models’ understanding. Both contributions have been quantitatively and qualitatively compared and assessed in a 5 to 10 times larger evaluation study on the ILSVRC2012 dataset against nine of the most recent and performing CAM techniques across six metrics. Expected Grad-CAM outperformed not only the original formulation but also more advanced methods, resulting in the second-best explainer with an Ins-Del score of 0.56. Hyper Expected Grad-CAM provided remarkable results across each quantitative metric, yielding a 0.15 increase in insertion when compared to the highest-scoring explainer PolyCAM, totaling to an Ins-Del score of 0.72.