Image Super resolution is a widely-studied problem in computer vision, where the objective is to convert a lowresolution image to a high resolution image. Conventional methods for achieving super-resolution such as image priors, interpolation, sparse coding require a lot of pre/post processing and optimization. Recently, deep learning methods such as convolutional neural networks and generative adversarial networks are being used to perform super-resolution with results competitive to the state of the art but none of them have been used on microscopy images. In this thesis, a generative adversarial network, mSRGAN, is proposed for super resolution with a perceptual loss function consisting of a adversarial loss, mean squared error and content loss. The objective of our implementation is to learn an end to end mapping between the low / high resolution images and optimize the upscaled image for quantitative metrics as well as perceptual quality. We then compare our results with the current state of the art methods in super resolution, conduct a proof of concept segmentation study to show that super resolved images can be used as a effective pre processing step before segmentation and validate the findings statistically. / Image Super-resolution är ett allmänt studerad problem i datasyn, där målet är att konvertera en lågupplösningsbild till en högupplöst bild. Konventionella metoder för att uppnå superupplösning som image priors, interpolation, sparse coding behöver mycket föroch efterbehandling och optimering.Nyligen djupa inlärningsmetoder som convolutional neurala nätverk och generativa adversariella nätverk är användas för att utföra superupplösning med resultat som är konkurrenskraftiga mot toppmoderna teknik, men ingen av dem har använts på mikroskopibilder. I denna avhandling, ett generativ kontradiktorisktsnätverk, mSRGAN, är föreslås för superupplösning med en perceptuell förlustfunktion bestående av en motsatt förlust, medelkvadratfel och innehållförlust.Mål med vår implementering är att lära oss ett slut på att slut kartläggning mellan bilder med låg / hög upplösning och optimera den uppskalade bilden för kvantitativa metriks såväl som perceptuell kvalitet. Vi jämför sedan våra resultat med de nuvarande toppmoderna metoderna i superupplösning, och uppträdande ett bevis på konceptsegmenteringsstudie för att visa att superlösa bilder kan användas som ett effektivt förbehandling steg före segmentering och validera fynden statistiskt.

Das Bereitstellen von Parametern bezüglich Auflösungsvermögen und effektiver Auflösung ist ein gut erforschtes Wissenschaftsfeld, dennoch sind noch einige offen Fragen zu klären, wenn eine standardisierte Erhebung angestrebt wird. Zu diesem Zweck ist im Rahmen der vorliegenden Arbeit ein Framework definiert und mathematisch und methodologisch beschrieben worden unter Einbeziehung aller untergeordneten Prozesse. Weiterhin liefert sie einen detaillierten Überblick zu den verwendeten Methoden und Strukturen, um räumliche Auflösung zu messen. Das zuvor definierte Framework wird darüber hinaus genutzt, um alle zugehörigen Probleme bezüglich eines genormten Prozesses zu identifizieren und zu lösen. Der so definierte Prozess ist außerdem Teil der bevorstehenden, neuen Norm: DIN 18740-8. Im Hinblick auf die Norm sind alle Messeinflüsse an den möglichen Stellen quantifiziert worden und an Stellen, wo dies nicht möglich ist, wurden Vorkehrungen definiert, die diese Einflüsse mindern. Darüber hinaus wurde ein zugehöriges Softwaretool entwickelt, das ebenfalls die neue Norm unterstützt. Als weiterer Schwerpunkt dieser Arbeit wurde ein Verfahren zur Verbesserung der räumlichen Auflösung entwickelt und bewertet. Das zugehörige Softwaretool kombiniert dabei verschiedene Super-Resolution-Ansätze unter Einbeziehung zusätzlicher Kenntnis über die Bildqualität. Der neuartige Super-Resolution-Ansatz verbessert die räumliche Auflösung von Luftbildern und True-Ortho-Mosaiken indem er ein Set von niedrig aufgelösten Rohbildern, deren optimierter, äußerer und innerer Orientierung und die abgeleitete 3D-Oberfläche als Eingangsdaten akzeptiert. Anschließend werden ein oder mehrere hochaufgelöste Bilder als hybride Kombination von klassischen Super-Resolution-Methoden und De-Mosaikierung berechnet, unter Berücksichtigung der photogrammetrischen Projektionen auf die 3D-Oberfläche. Dabei werden Limitierungen der Bildkoregistrierung mit üblich verwendeten Optical-Flow-Ansätzen überwunden. / Although acquisition of resolving power and effective spatial resolution is a well-studied field of research, there are still several scientific questions to be answered when it comes to a standardized determination. Therefore, this thesis provides a description of a framework for the imaging process of remote sensing sensors mathematically and methodologically including imaging components and subsequent processes. Furthermore, a detailed review for different structures and methods to measure spatial resolution is included. Aforementioned framework then is utilized to identify related issues to a standardized process obtaining spatial resolution parameters as an image quality criterion to support an upcoming standard DIN 18740-8. With respect to define the norm-procedure every measurement influence is quantified where possible and in other cases arrangements are specified to diminish their influence. Moreover, the development of an associated software measurement tool has been accomplished as part of this thesis, which also supports the norm for aerial image quality, spatial resolution in particular. As part of a further objective of this thesis, a super-resolution approach to improve spatial resolution of aerial images has been developed and evaluated. The related software tool is able to combine different super-resolution techniques and includes known image quality parameter in subsequent calculations. The novel super-resolution approach improves spatial resolution of aerial imagery and true ortho-mosaics by taking a set of multiple low-resolved raw images (color filter array), their optimized exterior and interior orientation parameters and the derived 3D-surface as input. Then, one or more super-resolved images are calculated as a hybrid of classic super-resolution method and demosaicing while considering photogrammetric back-projections onto the 3D-surface. Thereby, limitations of image co-registration with commonly used optical flow approaches can be neglected.

No description available.

Today’s 3D imaging techniques offer significant benefits over conventional 2D imaging techniques. The presence of natural depth information in the scene affords the observer an overall improved sense of reality and naturalness. A variety of systems attempting to reach this goal have been designed by many independent research groups, such as stereoscopic and auto-stereoscopic systems. Though the images displayed by such systems tend to cause eye strain, fatigue and headaches after prolonged viewing as users are required to focus on the screen plane/accommodation to converge their eyes to a point in space in a different plane/convergence. Holoscopy is a 3D technology that targets overcoming the above limitations of current 3D technology and was recently developed at Brunel University. This work is part W4.1 of the 3D VIVANT project that is funded by the EU under the ICT program and coordinated by Dr. Aman Aggoun at Brunel University, West London, UK. The objective of the work described in this thesis is to develop estimation and segmentation techniques that are capable of estimating precise 3D depth, and are applicable for holoscopic 3D imaging system. Particular emphasis is given to the task of automatic techniques i.e. favours algorithms with broad generalisation abilities, as no constraints are placed on the setting. Algorithms that provide invariance to most appearance based variation of objects in the scene (e.g. viewpoint changes, deformable objects, presence of noise and changes in lighting). Moreover, have the ability to estimate depth information from both types of holoscopic 3D images i.e. Unidirectional and Omni-directional which gives horizontal parallax and full parallax (vertical and horizontal), respectively. The main aim of this research is to develop 3D depth estimation and 3D image segmentation techniques with great precision. In particular, emphasis on automation of thresholding techniques and cues identifications for development of robust algorithms. A method for depth-through-disparity feature analysis has been built based on the existing correlation between the pixels at a one micro-lens pitch which has been exploited to extract the viewpoint images (VPIs). The corresponding displacement among the VPIs has been exploited to estimate the depth information map via setting and extracting reliable sets of local features. ii Feature-based-point and feature-based-edge are two novel automatic thresholding techniques for detecting and extracting features that have been used in this approach. These techniques offer a solution to the problem of setting and extracting reliable features automatically to improve the performance of the depth estimation related to the generalizations, speed and quality. Due to the resolution limitation of the extracted VPIs, obtaining an accurate 3D depth map is challenging. Therefore, sub-pixel shift and integration is a novel interpolation technique that has been used in this approach to generate super-resolution VPIs. By shift and integration of a set of up-sampled low resolution VPIs, the new information contained in each viewpoint is exploited to obtain a super resolution VPI. This produces a high resolution perspective VPI with wide Field Of View (FOV). This means that the holoscopic 3D image system can be converted into a multi-view 3D image pixel format. Both depth accuracy and a fast execution time have been achieved that improved the 3D depth map. For a 3D object to be recognized the related foreground regions and depth information map needs to be identified. Two novel unsupervised segmentation methods that generate interactive depth maps from single viewpoint segmentation were developed. Both techniques offer new improvements over the existing methods due to their simple use and being fully automatic; therefore, producing the 3D depth interactive map without human interaction. The final contribution is a performance evaluation, to provide an equitable measurement for the extent of the success of the proposed techniques for foreground object segmentation, 3D depth interactive map creation and the generation of 2D super-resolution viewpoint techniques. The no-reference image quality assessment metrics and their correlation with the human perception of quality are used with the help of human participants in a subjective manner.

This thesis is focused on understanding the interactions between molecules and surface-enhanced Raman scattering (SERS) substrates that are typically unresolved due to the diffraction limit of light. Towards this end, we have developed and tested several different sub-diffraction-limited imaging techniques in order to observe these interactions. First, we utilize an isotope-edited bianalyte approach combined with super-resolution imaging via Gaussian point-spread function fitting to elucidate the role of Raman reporter molecules on the location of the SERS emission centroids. By using low concentrations of two different analyte molecules, we find that the location of the SERS emission centroid depends on the number and positions of the molecules present on the SERS substrate. It is also known that SERS enhancement partially results from the molecule coupling its emission into the far-field through the plasmonic nanostructure. This results in a particle-dictated, dipole-like emission pattern, which cannot be accurately modeled as a Gaussian, so we tested the applicability of super-resolution imaging using a dipole-emission fitting model to this data. To test this model, we first fit gold nanorod (AuNR) luminescence images, as AuNR luminescence is primarily coupled out through the longitudinal dipole plasmon mode. This study showed that a three-dimensional dipole model is necessary to fit the AuNR emission, with the model providing accurate orientation and emission wavelength parameters for the nanostructure, as confirmed using correlated AFM and spectroscopy. The dipole fitting technique was next applied to single- and multiple-molecule SERS emission from silver nanoparticle dimers. We again found that a three-dimensional dipole PSF was necessary to accurately model the emission and orientation parameters of the dimer, but that at the single molecule level, the movement of the molecule causes increased uncertainty in the orientation parameters determined by the fit. Finally, we describe progress towards using a combined atomic force/optical microscope system in order to position a carbon nanotube analyte at known locations on the nanoparticle substrate. This would allow for the simultaneous mapping of nanoparticle topography and exact locations of plasmonic enhancement around the nanostructure, but consistently low signal-to-noise kept this technique from being viable. / text

<p>FtsZ, a bacterial tubulin homologue, is a cytoskeleton protein that plays key roles in cytokinesis of almost all prokaryotes. FtsZ assembles into protofilaments (pfs), one subunit thick, and these pfs assemble further to form a “Z ring” at the center of prokaryotic cells. The Z ring generates a constriction force on the inner membrane, and also serves as a scaffold to recruit cell-wall remodeling proteins for complete cell division in vivo. FtsZ can be subdivided into 3 main functional regions: globular domain, C terminal (Ct) linker, and Ct peptide. The globular domain binds GTP to assembles the pfs. The extreme Ct peptide binds membrane proteins to allow cytoplasmic FtsZ to function at the inner membrane. The Ct linker connects the globular domain and Ct peptide. In the present studies, we used genetic and structural approaches to investigate the function of Escherichia coli (E. coli) FtsZ. We sought to examine three questions: (1) Are lateral bonds between pfs essential for the Z ring? (2) Can we improve direct visualization of FtsZ in vivo by engineering an FtsZ-FP fusion that can function as the sole source of FtsZ for cell division? (3) Is the divergent Ct linker of FtsZ an intrinsically disordered peptide (IDP)?</p><p> One model of the Z ring proposes that pfs associate via lateral bonds to form ribbons; however, lateral bonds are still only hypothetical. To explore potential lateral bonding sites, we probed the surface of E. coli FtsZ by inserting either small peptides or whole FPs. Of the four lateral surfaces on FtsZ pfs, we obtained inserts on the front and back surfaces that were functional for cell division. We concluded that these faces are not sites of essential interactions. Inserts at two sites, G124 and R174 located on the left and right surfaces, completely blocked function, and were identified as possible sites for essential lateral interactions. Another goal was to find a location within FtsZ that supported fusion of FP reporter proteins, while allowing the FtsZ-FP to function as the sole source of FtsZ. We discovered one internal site, G55-Q56, where several different FPs could be inserted without impairing function. These FtsZ-FPs may provide advances for imaging Z-ring structure by super-resolution techniques.</p><p> The Ct linker is the most divergent region of FtsZ in both sequence and length. In E. coli FtsZ the Ct linker is 50 amino acids (aa), but for other FtsZ it can be as short as 37 aa or as long as 250 aa. The Ct linker has been hypothesized to be an IDP. In the present study, circular dichroism confirmed that isolated Ct linkers of E. coli (50 aa) and C. crescentus (175 aa) are IDPs. Limited trypsin proteolysis followed by mass spectrometry (LC-MS/MS) confirmed Ct linkers of E. coli (50 aa) and B. subtilis (47 aa) as IDPs even when still attached to the globular domain. In addition, we made chimeras, swapping the E. coli Ct linker for other peptides and proteins. Most chimeras allowed for normal cell division in E. coli, suggesting that IDPs with a length of 43 to 95 aa are tolerated, sequence has little importance, and electrostatic charge is unimportant. Several chimeras were purified to confirm the effect they had on pf assembly. We concluded that the Ct linker functions as a flexible tether allowing for force to be transferred from the FtsZ pf to the membrane to constrict the septum for division.</p> / Dissertation

No description available.

Multi-frame image super-resolution is a procedure which takes several noisy low-resolution images of the same scene, acquired under different conditions, and processes them together to synthesize one or more high-quality super-resolution images, with higher spatial frequency, and less noise and image blur than any of the original images. The inputs can take the form of medical images, surveillance footage, digital video, satellite terrain imagery, or images from many other sources. This thesis focuses on Bayesian methods for multi-frame super-resolution, which use a prior distribution over the super-resolution image. The goal is to produce outputs which are as accurate as possible, and this is achieved through three novel super-resolution schemes presented in this thesis. Previous approaches obtained the super-resolution estimate by first computing and fixing the imaging parameters (such as image registration), and then computing the super-resolution image with this registration. In the first of the approaches taken here, superior results are obtained by optimizing over both the registrations and image pixels, creating a complete simultaneous algorithm. Additionally, parameters for the prior distribution are learnt automatically from data, rather than being set by trial and error. In the second approach, uncertainty in the values of the imaging parameters is dealt with by marginalization. In a previous Bayesian image super-resolution approach, the marginalization was over the super-resolution image, necessitating the use of an unfavorable image prior. By integrating over the imaging parameters rather than the image, the novel method presented here allows for more realistic prior distributions, and also reduces the dimension of the integral considerably, removing the main computational bottleneck of the other algorithm. Finally, a domain-specific image prior, based upon patches sampled from other images, is presented. For certain types of super-resolution problems where it is applicable, this sample-based prior gives a significant improvement in the super-resolution image quality.

No description available.

No description available.