During software development and maintenance, as a software system evolves, changes are made and bugs are fixed in various files. In large-scale systems, file histories are stored in software repositories, such as CVS, which record modifications. By studying software repositories, we can learn about open source software development rocesses. Knowing where these changes will happen in advance, gives power to managers and developers to concentrate on those files. Due to the unpredictability in software development process, proposing an accurate change prediction model is hard. It is even harder to compare different models with the actual model of changes that is not available. <br /><br /> In this thesis, we first analyze the information generated during the development process, which can be obtained through mining the software repositories. We observe that the change data follows a Zipf distribution and exhibits self-similarity. Based on the extracted data, we then develop three probabilistic models to predict which files will have changes or bugs. One purpose of creating these models is to rank the files of the software that are most susceptible to having faults. <br /><br /> The first model is Maximum Likelihood Estimation (MLE), which simply counts the number of events i. e. , changes or bugs that occur in to each file, and normalizes the counts to compute a probability distribution. The second model is Reflexive Exponential Decay (RED), in which we postulate that the predictive rate of modification in a file is incremented by any modification to that file and decays exponentially. The result of a new bug occurring to that file is a new exponential effect added to the first one. The third model is called RED Co-Changes (REDCC). With each modification to a given file, the REDCC model not only increments its predictive rate, but also increments the rate for other files that are related to the given file through previous co-changes. <br /><br /> We then present an information-theoretic approach to evaluate the performance of different prediction models. In this approach, the closeness of model distribution to the actual unknown probability distribution of the system is measured using cross entropy. We evaluate our prediction models empirically using the proposed information-theoretic approach for six large open source systems. Based on this evaluation, we observe that of our three prediction models, the REDCC model predicts the distribution that is closest to the actual distribution for all the studied systems.

During software development and maintenance, as a software system evolves, changes are made and bugs are fixed in various files. In large-scale systems, file histories are stored in software repositories, such as CVS, which record modifications. By studying software repositories, we can learn about open source software development rocesses. Knowing where these changes will happen in advance, gives power to managers and developers to concentrate on those files. Due to the unpredictability in software development process, proposing an accurate change prediction model is hard. It is even harder to compare different models with the actual model of changes that is not available. <br /><br /> In this thesis, we first analyze the information generated during the development process, which can be obtained through mining the software repositories. We observe that the change data follows a Zipf distribution and exhibits self-similarity. Based on the extracted data, we then develop three probabilistic models to predict which files will have changes or bugs. One purpose of creating these models is to rank the files of the software that are most susceptible to having faults. <br /><br /> The first model is Maximum Likelihood Estimation (MLE), which simply counts the number of events i. e. , changes or bugs that occur in to each file, and normalizes the counts to compute a probability distribution. The second model is Reflexive Exponential Decay (RED), in which we postulate that the predictive rate of modification in a file is incremented by any modification to that file and decays exponentially. The result of a new bug occurring to that file is a new exponential effect added to the first one. The third model is called RED Co-Changes (REDCC). With each modification to a given file, the REDCC model not only increments its predictive rate, but also increments the rate for other files that are related to the given file through previous co-changes. <br /><br /> We then present an information-theoretic approach to evaluate the performance of different prediction models. In this approach, the closeness of model distribution to the actual unknown probability distribution of the system is measured using cross entropy. We evaluate our prediction models empirically using the proposed information-theoretic approach for six large open source systems. Based on this evaluation, we observe that of our three prediction models, the REDCC model predicts the distribution that is closest to the actual distribution for all the studied systems.

Superpixels are the results of an image oversegmentation. They are an established intermediate level image representation and used for various applications including object detection, 3d reconstruction and semantic segmentation. While there are various approaches to create such segmentations, there is a lack of knowledge about their properties. In particular, there are contradicting results published in the literature. This thesis identifies segmentation quality, stability, compactness and runtime to be important properties of superpixel segmentation algorithms. While for some of these properties there are established evaluation methodologies available, this is not the case for segmentation stability and compactness. Therefore, this thesis presents two novel metrics for their evaluation based on ground truth optical flow. These two metrics are used together with other novel and existing measures to create a standardized benchmark for superpixel algorithms. This benchmark is used for extensive comparison of available algorithms. The evaluation results motivate two novel segmentation algorithms that better balance trade-offs of existing algorithms: The proposed Preemptive SLIC algorithm incorporates a local preemption criterion in the established SLIC algorithm and saves about 80 % of the runtime. The proposed Compact Watershed algorithm combines Seeded Watershed segmentation with compactness constraints to create regularly shaped, compact superpixels at the even higher speed of the plain watershed transformation. Operating autonomous systems over the course of days, weeks or months, based on visual navigation, requires repeated recognition of places despite severe appearance changes as they are for example induced by illumination changes, day-night cycles, changing weather or seasons - a severe problem for existing methods. Therefore, the second part of this thesis presents two novel approaches that incorporate superpixel segmentations in place recognition in changing environments. The first novel approach is the learning of systematic appearance changes. Instead of matching images between, for example, summer and winter directly, an additional prediction step is proposed. Based on superpixel vocabularies, a predicted image is generated that shows, how the summer scene could look like in winter or vice versa. The presented results show that, if certain assumptions on the appearance changes and the available training data are met, existing holistic place recognition approaches can benefit from this additional prediction step. Holistic approaches to place recognition are known to fail in presence of viewpoint changes. Therefore, this thesis presents a new place recognition system based on local landmarks and Star-Hough. Star-Hough is a novel approach to incorporate the spatial arrangement of local image features in the computation of image similarities. It is based on star graph models and Hough voting and particularly suited for local features with low spatial precision and high outlier rates as they are expected in the presence of appearance changes. The novel landmarks are a combination of local region detectors and descriptors based on convolutional neural networks. This thesis presents and evaluates several new approaches to incorporate superpixel segmentations in local region detection. While the proposed system can be used with different types of local regions, in particular the combination with regions obtained from the novel multiscale superpixel grid shows to perform superior to the state of the art methods - a promising basis for practical applications.

Superpixels are the results of an image oversegmentation. They are an established intermediate level image representation and used for various applications including object detection, 3d reconstruction and semantic segmentation. While there are various approaches to create such segmentations, there is a lack of knowledge about their properties. In particular, there are contradicting results published in the literature. This thesis identifies segmentation quality, stability, compactness and runtime to be important properties of superpixel segmentation algorithms. While for some of these properties there are established evaluation methodologies available, this is not the case for segmentation stability and compactness. Therefore, this thesis presents two novel metrics for their evaluation based on ground truth optical flow. These two metrics are used together with other novel and existing measures to create a standardized benchmark for superpixel algorithms. This benchmark is used for extensive comparison of available algorithms. The evaluation results motivate two novel segmentation algorithms that better balance trade-offs of existing algorithms: The proposed Preemptive SLIC algorithm incorporates a local preemption criterion in the established SLIC algorithm and saves about 80 % of the runtime. The proposed Compact Watershed algorithm combines Seeded Watershed segmentation with compactness constraints to create regularly shaped, compact superpixels at the even higher speed of the plain watershed transformation. Operating autonomous systems over the course of days, weeks or months, based on visual navigation, requires repeated recognition of places despite severe appearance changes as they are for example induced by illumination changes, day-night cycles, changing weather or seasons - a severe problem for existing methods. Therefore, the second part of this thesis presents two novel approaches that incorporate superpixel segmentations in place recognition in changing environments. The first novel approach is the learning of systematic appearance changes. Instead of matching images between, for example, summer and winter directly, an additional prediction step is proposed. Based on superpixel vocabularies, a predicted image is generated that shows, how the summer scene could look like in winter or vice versa. The presented results show that, if certain assumptions on the appearance changes and the available training data are met, existing holistic place recognition approaches can benefit from this additional prediction step. Holistic approaches to place recognition are known to fail in presence of viewpoint changes. Therefore, this thesis presents a new place recognition system based on local landmarks and Star-Hough. Star-Hough is a novel approach to incorporate the spatial arrangement of local image features in the computation of image similarities. It is based on star graph models and Hough voting and particularly suited for local features with low spatial precision and high outlier rates as they are expected in the presence of appearance changes. The novel landmarks are a combination of local region detectors and descriptors based on convolutional neural networks. This thesis presents and evaluates several new approaches to incorporate superpixel segmentations in local region detection. While the proposed system can be used with different types of local regions, in particular the combination with regions obtained from the novel multiscale superpixel grid shows to perform superior to the state of the art methods - a promising basis for practical applications.

Une nouvelle approche du traitement de l'image, appelée SOM-QE, qui exploite quantization error (QE) des self-organizing maps (SOM) est proposée dans cette thèse. Les SOM produisent des représentations discrètes de faible dimension des données d'entrée de haute dimension. QE est déterminée à partir des résultats du processus d'apprentissage non supervisé du SOM et des données d'entrée. SOM-QE d'une série chronologique d'images peut être utilisé comme indicateur de changements dans la série chronologique. Pour configurer SOM, on détermine la taille de la carte, la distance du voisinage, le rythme d'apprentissage et le nombre d'itérations dans le processus d'apprentissage. La combinaison de ces paramètres, qui donne la valeur la plus faible de QE, est considérée comme le jeu de paramètres optimal et est utilisée pour transformer l'ensemble de données. C'est l'utilisation de l'assouplissement quantitatif. La nouveauté de la technique SOM-QE est quadruple : d'abord dans l'usage. SOM-QE utilise un SOM pour déterminer la QE de différentes images - typiquement, dans un ensemble de données de séries temporelles - contrairement à l'utilisation traditionnelle où différents SOMs sont appliqués sur un ensemble de données. Deuxièmement, la valeur SOM-QE est introduite pour mesurer l'uniformité de l'image. Troisièmement, la valeur SOM-QE devient une étiquette spéciale et unique pour l'image dans l'ensemble de données et quatrièmement, cette étiquette est utilisée pour suivre les changements qui se produisent dans les images suivantes de la même scène. Ainsi, SOM-QE fournit une mesure des variations à l'intérieur de l'image à une instance dans le temps, et lorsqu'il est comparé aux valeurs des images subséquentes de la même scène, il révèle une visualisation transitoire des changements dans la scène à l'étude. Dans cette recherche, l'approche a été appliquée à l'imagerie artificielle, médicale et géographique pour démontrer sa performance. Les scientifiques et les ingénieurs s'intéressent aux changements qui se produisent dans les scènes géographiques d'intérêt, comme la construction de nouveaux bâtiments dans une ville ou le recul des lésions dans les images médicales. La technique SOM-QE offre un nouveau moyen de détection automatique de la croissance dans les espaces urbains ou de la progression des maladies, fournissant des informations opportunes pour une planification ou un traitement approprié. Dans ce travail, il est démontré que SOM-QE peut capturer de très petits changements dans les images. Les résultats confirment également qu'il est rapide et moins coûteux de faire la distinction entre le contenu modifié et le contenu inchangé dans les grands ensembles de données d'images. La corrélation de Pearson a confirmé qu'il y avait des corrélations statistiquement significatives entre les valeurs SOM-QE et les données réelles de vérité de terrain. Sur le plan de l'évaluation, cette technique a donné de meilleurs résultats que les autres approches existantes. Ce travail est important car il introduit une nouvelle façon d'envisager la détection rapide et automatique des changements, même lorsqu'il s'agit de petits changements locaux dans les images. Il introduit également une nouvelle méthode de détermination de QE, et les données qu'il génère peuvent être utilisées pour prédire les changements dans un ensemble de données de séries chronologiques. / A new approach for image processing, dubbed SOM-QE, that exploits the quantization error (QE) from self-organizing maps (SOM) is proposed in this thesis. SOM produce low-dimensional discrete representations of high-dimensional input data. QE is determined from the results of the unsupervised learning process of SOM and the input data. SOM-QE from a time-series of images can be used as an indicator of changes in the time series. To set-up SOM, a map size, the neighbourhood distance, the learning rate and the number of iterations in the learning process are determined. The combination of these parameters that gives the lowest value of QE, is taken to be the optimal parameter set and it is used to transform the dataset. This has been the use of QE. The novelty in SOM-QE technique is fourfold: first, in the usage. SOM-QE employs a SOM to determine QE for different images - typically, in a time series dataset - unlike the traditional usage where different SOMs are applied on one dataset. Secondly, the SOM-QE value is introduced as a measure of uniformity within the image. Thirdly, the SOM-QE value becomes a special, unique label for the image within the dataset and fourthly, this label is used to track changes that occur in subsequent images of the same scene. Thus, SOM-QE provides a measure of variations within the image at an instance in time, and when compared with the values from subsequent images of the same scene, it reveals a transient visualization of changes in the scene of study. In this research the approach was applied to artificial, medical and geographic imagery to demonstrate its performance. Changes that occur in geographic scenes of interest, such as new buildings being put up in a city or lesions receding in medical images are of interest to scientists and engineers. The SOM-QE technique provides a new way for automatic detection of growth in urban spaces or the progressions of diseases, giving timely information for appropriate planning or treatment. In this work, it is demonstrated that SOM-QE can capture very small changes in images. Results also confirm it to be fast and less computationally expensive in discriminating between changed and unchanged contents in large image datasets. Pearson's correlation confirmed that there was statistically significant correlations between SOM-QE values and the actual ground truth data. On evaluation, this technique performed better compared to other existing approaches. This work is important as it introduces a new way of looking at fast, automatic change detection even when dealing with small local changes within images. It also introduces a new method of determining QE, and the data it generates can be used to predict changes in a time series dataset.