The usage of deep learning algorithms such as Convolutional Neural Networks within the field of medical imaging has grown in popularity over the past few years. In particular, these types of algorithms have been used to detect abnormalities in chest x-rays, one of the most commonly performed type of radiographic examination. To try and improve the workflow of radiologists, this thesis investigated the possibility of using convolutional neural networks to create a lung triage to sort a bulk of chest x-ray images based on a degree of disease, where sick lungs should be prioritized before healthy lungs. The results from using a binary relevance approach to train multiple classifiers for different observations commonly found in chest x-rays shows that several models fail to learn how to classify x-ray images, most likely due to insufficient and/or imbalanced data. Using a binary relevance approach to create a triage is feasible but inflexible due to having to handle multiple models simultaneously. In future work it would therefore be interesting to further investigate other approaches, such as a single binary classification model or a multi-label classification model.

Labeling problem, due to its versatile modeling ability, is widely used in various image analysis tasks. In practice, certain prior information is often available to be embedded in the model to increase accuracy and robustness. However, it is not always straightforward to formulate the problem so that the prior information is correctly incorporated. It is even more challenging that the proposed model admits efficient algorithms to find globally optimal solution. In this dissertation, a series of natural and medical image segmentation tasks are modeled as labeling problems. Each proposed model incorporates different useful prior information. These prior information includes ordering constraints between certain labels, soft user input enforcement, multi-scale context between over-segmented regions and original voxel, multi-modality context prior, location context between multiple modalities, star-shape prior, and gradient vector flow shape prior. With judicious exploitation of each problem's intricate structure, efficient and exact algorithms are designed for all proposed models. The efficient computation allow the proposed models to be applied on large natural and medical image datasets using small memory footprint and reasonable time assumption. The global optimality guarantee makes the methods robust to local noise and easy to debug. The proposed models and algorithms are validated on multiple experiments, using both natural and medical images. Promising and competitive results are shown when compared to state-of-art.

[pt] Este trabalho consiste na caracterização da fase gama linha em superligas de níquel através do processamento e análise digital de imagens. Amostras de uma superliga de níquel foram submetidas a 10 tratamentos térmicos diferentes. Após o tratamento térmico, estas amostras foram preparadas para avaliação metalográfica e imagens da microestrutura foram adquiridas no microscópio eletrônico de varredura (MEV). A fase gama linha presente no material foi segmentada e posteriormente medida através do processamento digital de imagens. Foram analisadas 429 imagens, o que gerou medidas de mais de 225,000 partículas. Foi medida a fração de área ocupada pelas partículas de gama linha, a quantidade de partículas por imagem, a área de cada partícula e a razão de aspectos das partículas. Uma análise estatística dos resultados foi realizada. Os resultados mostraram menor concentração e maior tamanho de gama linha com o aumento da temperatura, conforme era esperado. O uso de análise digital de imagens permitiu obter medidas com altíssima confiabilidade, uma vez que a alta velocidade de análise permite uma amostragem muito superior às técnicas tradicionais de caracterização microestrutural. / [en] This study consists of the characterization of the gamma prime phase in nickel base superalloys using digital image processing and analysis. Samples of a nickel base superalloy were heat treated to 10 different temperatures. After the heat treatment cycle, the samples were prepared for metallographic evaluation and images of the microstructure were acquired using a scanning electron microscope (SEM). The gamma prime was segmented and measured using digital image processing methods. 429 images were analyzed, in which over 225000 particles were measured. The following parameters were obtained: area fraction of gamma prime, number of particles of gamma prime per image, area of each particle and the particles' feret ratio. An statistical analysis of the data was performed The results showed that the gamma prime fraction decreases and its size increases with temperature increase, as expected. The use of digital image analysis provided measurements with high accuracy, as the high speed of digital image analysis allowed sampling much superior to the traditional microstructural characterization methods.

The prepartum surge in fetal plasma cortisol, essential for the maturation of organs in mammals and the normal timing of parturition in some species, including sheep, may result from an increase in the molar ratio of adrenocorticotropin (ACTH) to pro-opiomelanocortin (POMC) in the fetal circulation. Related to this, the cleavage of POMC to ACTH by the enzyme, prohormone convertase 1 (PC1), may be influenced by corticotrophin releasing hormone (CRH) stimulation. Accumulating evidence suggests that the capacity of individual corticotrophs to process POMC to ACTH may vary and individual corticotrophs are differentially responsive to CRH. It is not known, however, if there are separate corticotroph subpopulations in the fetal sheep pituitary which can be identified by differential colocalisation of POMC, ACTH and the CRH receptor 1, CRHR₁, nor if changes in the relative proportions of such subpopulations play a role in the molecular mechanisms underlying the overall changes in pituitary function described previously during gestation and in response to suboptimal uterine environments. To investigate these hypotheses, it was first necessary to develop novel methods for the simultaneous immunohistochemical labelling of POMC, ACTH and CRHR₁ in individual cells on sections of fetal sheep pituitary. In addition, I developed and validated an automated method to categorise and count individual cells to increase the quantitative power of this study. Pituitary tissue was collected from control fetuses at 53-55 (n=6), 63-85 (n=6), 110 (n=4), 139-141 (n=4) and 144-145 (n=6) days gestation. Two animal models, known to alter pituitary function in the fetal sheep, were used to investigate corticotrophic adaptations to suboptimal uterine environments. For the maternal periconceptional undernutrition (PCUN) model, maternal feed was reduced to 70% of maintenance requirements from at least 45 days before to 7 days after mating and fetal tissues were collected at 53-55 days gestation (n=7). For the placental restriction (PR) model, the majority of the placental attachment sites were removed in five ewes before mating and fetal tissues were collected at 140 (n=4) and 144 (n=4) days gestation. Pituitary sections were simultaneously labelled with antisera raised against full length POMC, ACTH and CRHR₁ and the proportions of pituitary cells with combinations of antisera were quantified. Four subpopulations of corticotrophs were identified, which expressed either: POMC+ACTH+CRHR₁, ACTH+CRHR₁, POMC+ CRHR₁ or POMConly. There was a significant decrease in the proportion of pituitary cells expressing POMC+ACTH+CRHR₁ between 53-55 and 65-85 days gestation, before an increase at 110 days gestation and a further marked decrease between 139-141 and 144-145 days gestation. In fetuses from the PCUN group, the proportion of pituitary cells expressing POMC+ACTH+CRHR₁ in early gestation was reduced. PR resulted in a significantly higher proportion of corticotrophs expressing POMC+ACTH+CRHR₁ during the prepartum period. This work represents the discovery of the differential expression of POMC, ACTH and CRHR₁ in individual corticotrophs of the fetal sheep pituitary and the first insights into the pituitary adaptations to periconceptional nutrient restriction and placental restriction at the level of individual corticotrophs. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=1337370 / Thesis (Ph.D.) -- University of Adelaide, School of Molecular and Biomedical Science, 2008

<p>The market for unmanned aerial vehicles (UAVs) is growing rapidly. In order to meet the demand for marine applications CybAero AB has recently started a project named Mobile Automatic Launch and Landing Station (MALLS). MALLS enables the uav to land on moving targets such as ships. This thesis studies a system that estimates the pose of a helicopter in order to land on a moving target.</p><p>The main focus has been on developing a pose estimation system using computer vision. Two different methods for estimating the pose have been studied, one using homography and one using an Extended Kalman Filter (ekf). Both methods have been tested on real flight data from a camera mounted on a RC-helicopter. The accuracy of the pose estimation system has been verified with data from a test with the camera mounted on an industrial robot. The test results show that the ekf-based system is less sensitive to noise than the homography-based system. The ekf-based system however requires initial values which the homography-based system does not. The accuracy of both systems is found to be good enough for the purpose.</p><p>A novel control system with control rules for performing an autonomous landing on a moving target has been developed. The control system has not been tested during flight.</p> / <p>Marknaden for obemannade autonoma luftburna farkoster (UAV:er) växer snabbt. För att möta behovet av marina tillämpningar har CybAero AB nyligen startat ett projekt som kallas Mobil Automatisk Start- och Landningsstation (MALLS). Syftet med malls är att möjliggöra autonom start och landning på objekt i rörelse, som till exempel ett fartyg. I det här examensarbetet studeras ett system för att bestämma position och attityd för en helikopter relativt en helikopterplatta, för att möjliggöra landning på ett ojekt i rörelse.</p><p>Fokus har främst legat på att utveckla ett positionerings- och attitydbestämningssystem. Ett datorseende positionerings- och attitydbestämningssystem har utvecklats. Två olika metoder har undersökts, ett system som bygger på homografi och ett annat som bygger på Extended Kalman Filter (EKF). Båda metoderna har testats med verklig data från en kamera monterad på en RC helikopter. Noggrannheten i positionsbestämmelsen har undersökts med hjälp av data från en industrirobot. Testresultaten visar att det EKF-baserade systemet är mindre bruskänsligt än det homografibaserade systemet. En nackdel med det ekf-baserade systemet är däremot att det kräver initialvillkor vilket det homografibaserade systemet inte gör. Noggrannheten på båda systemen finner vi tillfredsställande för syftet.</p><p>Ett enkelt styrsystem med styrlagar för att genomföra landningar på ett rörligtobjekt har utvecklats. Styrsystemet har dock inte testats under verklig flygning.</p>

<p>Just how far is it possible to make learning of new parts for recognition and robot picking autonomous? This thesis initially gives the prerequisites for the steps in learning and calibration that are to be automated. Among these tasks are to select a suitable part model from numerous candidates with the help of a new part segmenter, as well as computing the spatial extent of this part, facilitating robotic collision handling. Other tasks are to analyze the part model in order to highlight correct and suitable edge segments for increasing pattern matching certainty, and to choose appropriate acceptance levels for pattern matching. Furthermore, tasks deal with simplifying camera calibration by analyzing the calibration pattern, as well as compensating for differences in perspective at great depth variations, by calculating the centre of perspective of the image. The image processing algorithms created in order to solve the tasks are described and evaluated thoroughly. This thesis shows that simplification of steps of learning and calibration, by the help of advanced image processing, really is possible.</p>

<p>Most people are familiar with the BRIO labyrinth game and the challenge of guiding the ball through the maze. The goal of this project was to use this game to create a platform for evaluation of control algorithms. The platform was used to evaluate a few different controlling algorithms, both traditional automatic control algorithms as well as algorithms based on online incremental learning.</p><p>The game was fitted with servo actuators for tilting the maze. A camera together with computer vision algorithms were used to estimate the state of the game. The evaluated controlling algorithm had the task of calculating a proper control signal, given the estimated state of the game.</p><p>The evaluated learning systems used traditional control algorithms to provide initial training data. After initial training, the systems learned from their own actions and after a while they outperformed the controller used to provide initial training.</p>

<p>An image that has been subject to the out-of-focus phenomenon has reducedsharpness, contrast and level of detail depending on the amount of defocus. Torestore out-of-focused images is a complex task due to the information loss thatoccurs. However there exist many restoration algorithms that attempt to revertthis defocus by estimating a noise model and utilizing the point spread function.The purpose of this thesis, proposed by FLIR Systems, was to find a robustalgorithm that can restore focus and from the customer’s perspective be userfriendly. The thesis includes three implemented algorithms that have been com-pared to MATLABs built-in. Three image series were used to evaluate the limitsand performance of each algorithm, based on deblurring quality, implementationcomplexity, computation time and usability.Results show that the Alternating Direction Method for total variation de-convolution proposed by Tao et al. [29] together with its the modified discretecosines transform version restores the defocused images with the highest qual-ity. These two algorithms include features such as, fast computational time, fewparameters to tune and a powerful noise reduction.</p>

<p>In this thesis work a part of the video coding standard H.264 has been implemented. The part of the video coder that is used to code the I-frames has been implemented to see how well suited it is for regular image coding. The big difference versus other image coding standards, such as JPEG and JPEG2000, is that this video coder uses both a predictor and a transform to compress the I-frames, while JPEG and JPEG2000 only use a transform. Since the prediction error is sent instead of the actual pixel values, a lot of the values are zero or close to zero before the transformation and quantization. The method is much like a video encoder but the difference is that blocks of an image are predicted instead of frames in a video sequence.</p> / <p>I det här examensarbetet har en del av videokodningsstandarden H.264 implementerats. Den del av videokodaren som används för att koda s.k. I-bilder har implementerats för att testa hur bra den fungerar för ren stillbildskodning. Den stora skillnaden mot andra stillbildskodningsmetoder, såsom JPEG och JPEG2000, är att denna videokodaren använder både en prediktor och en transform för att komprimera stillbilderna, till skillnad från JPEG och JPEG2000 som bara använder en transform. Eftersom prediktionsfelen skickas istället för själva pixelvärdena så är många värden lika med noll eller nära noll redan innan transformationen och kvantiseringen. Metoden liknar alltså till mycket en ren videokodare, med skillnaden att man predikterar block i en bild istället för bilder i en videosekvens.</p>

<p>Today, tool center point calibration is mostly done by a manual procedure. The method is very time consuming and the result may vary due to how skilled the operators are.</p><p>This thesis proposes a new automated iterative method for tool center point calibration of industrial robots, by making use of computer vision and image processing techniques. The new method has several advantages over the manual calibration method. Experimental verifications have shown that the proposed method is much faster, still delivering a comparable or even better accuracy. The setup of the proposed method is very easy, only one USB camera connected to a laptop computer is needed and no contact with the robot tool is necessary during the calibration procedure.</p><p>The method can be split into three different parts. Initially, the transformation between the robot wrist and the tool is determined by solving a closed loop of homogeneous transformations. Second an image segmentation procedure is described for finding point correspondences on a rotation symmetric robot tool. The image segmentation part is necessary for performing a measurement with six degrees of freedom of the camera to tool transformation. The last part of the proposed method is an iterative procedure which automates an ordinary four point tool center point calibration algorithm. The iterative procedure ensures that the accuracy of the tool center point calibration only depends on the accuracy of the camera when registering a movement between two positions.</p>