Automatic machine understanding of documents from image inputs enables many applications in modern document workflows, digital archives of historical documents, and general machine intelligence, among others. Together, the techniques for understanding document images comprise the field of Document Image Analysis (DIA). Within DIA, the research community has identified several sub-problems, such as page segmentation and Optical Character Recognition (OCR). As the field has matured, there has been a trend of moving away from heuristic-based methods, designed for particular tasks and domains of documents, and moving towards machine learning methods that learn to solve tasks from examples of input/output pairs. Within machine learning, a particular class of models, known as deep learning models, have established themselves as the state-of-the-art for many image-based applications, including DIA. While traditional machine learning models typically operate on features designed by researchers, deep learning models are able to learn task-specific features directly from raw pixel inputs.This dissertation is collection of papers that proposes several deep learning models to solve a variety of tasks within DIA. The first task is historical document binarization, where an input image of a degraded historical document is converted to a bi-tonal image to separate foreground text from background regions. The next part of the dissertation considers document segmentation problems, including identifying the boundary between the document page and its background, as well as segmenting an image of a data table into rows, columns, and cells. Finally, a variety of deep models are proposed to solve recognition tasks. These tasks include whole document image classification, identifying the font of a given piece of text, and transcribing handwritten text in low-resource languages.

The thesis deals with the type of font of hand writing which is taught at the elementary schools it the Czech Republic its main symbols and methods of the font rating. In the theoretical part I define the basic concepts, introduces the history, development of writing, actual methods of teaching of hand writing at the elementary schools and the overview of exercise books for practicing of the very first hand writing. In the practical part, I find out using a questionnaire according to which criteria the teachers evaluate children's hand writing. The research part contains written work of children. I analyze the texts where I evaluate qualitative features (shape, size, gradient, binding, modification) of bound and unbound font of writing in the third grade of elementary school. To improve writing skills, I suggest specific didactic utilities, release exercises, act. Key words: The bound font of writing, analytical-synthetic method, The unbound font of writing, genetic method, global method, Sfumato, font rating, qualitative writing symbols, exercise book, release exercises

Arabic text recognition was not researched as thoroughly as other natural languages. The need for automatic Arabic text recognition is clear. In addition to the traditional applications like postal address reading, check verification in banks, and office automation, there is a large interest in searching scanned documents that are available on the internet and for searching handwritten manuscripts. Other possible applications are building digital libraries, recognizing text on digitized maps, recognizing vehicle license plates, using it as first phase in text readers for visually impaired people and understanding filled forms. This research work aims to contribute to the current research in the field of optical character recognition (OCR) of printed Arabic text by developing novel techniques and schemes to advance the performance of the state of the art Arabic OCR systems. Statistical and analytical analysis for Arabic Text was carried out to estimate the probabilities of occurrences of Arabic character for use with Hidden Markov models (HMM) and other techniques. Since there is no publicly available dataset for printed Arabic text for recognition purposes it was decided to create one. In addition, a minimal Arabic script is proposed. The proposed script contains all basic shapes of Arabic letters. The script provides efficient representation for Arabic text in terms of effort and time. Based on the success of using HMM for speech and text recognition, the use of HMM for the automatic recognition of Arabic text was investigated. The HMM technique adapts to noise and font variations and does not require word or character segmentation of Arabic line images. In the feature extraction phase, experiments were conducted with a number of different features to investigate their suitability for HMM. Finally, a novel set of features, which resulted in high recognition rates for different fonts, was selected. The developed techniques do not need word or character segmentation before the classification phase as segmentation is a byproduct of recognition. This seems to be the most advantageous feature of using HMM for Arabic text as segmentation tends to produce errors which are usually propagated to the classification phase. Eight different Arabic fonts were used in the classification phase. The recognition rates were in the range from 98% to 99.9% depending on the used fonts. As far as we know, these are new results in their context. Moreover, the proposed technique could be used for other languages. A proof-of-concept experiment was conducted on English characters with a recognition rate of 98.9% using the same HMM setup. The same techniques where conducted on Bangla characters with a recognition rate above 95%. Moreover, the recognition of printed Arabic text with multi-fonts was also conducted using the same technique. Fonts were categorized into different groups. New high recognition results were achieved. To enhance the recognition rate further, a post-processing module was developed to correct the OCR output through character level post-processing and word level post-processing. The use of this module increased the accuracy of the recognition rate by more than 1%.

Cette monographie de géologie régionale concerne essentiellement la zone briançonnaise et effleure la zone du flysch . La stratigraphie de ce massif entre Guil et Ubaye est particuliérement etudiée. Des explications tectoniques sont apportéees . Ce travail date de 1935.

This diploma thesis focuses on the important knowledge related to education and bringing up left-handed children. The theoretical part focuses on the explanation of the term laterality and its diagnostics and also on the methodology of left-handed writing and the preparatory exercises for writing. Furthermore this part points to some various causes and effects suppression of left- handedness, and a brief insight into a former approach and an actual approach. This part offers information to the new trend in writing - Czech school font called Comenia Script. The practical part concentrates on the development of writing for left-handed in the commonly used writing alphabet, and in the font Comenia Script. There is also included a tentative questionnaire for primary school teachers and an observation and teaching of left- handed pupils in four different grades at primary schools. Keywords: laterality, left-handedness, upbringing, education, the commonly used writing alphabet, the font Comenia Script

According to the HTTP Archive, 75% of websites are using web fonts. Multiple conditions have to be met before modern web browsers like Chrome, Firefox and Safari decide to download the web fonts needed on a page. As a result, web fonts are late discovered resources that can delay the First Meaningful Paint (FMP). Improving the FMP is relevant for the web industry, particularly for performance-conscious web developers. This paper gives insight into how the resource prioritization approaches HTTP/2 Preload and HTTP/2 Server Push can be used to optimize the delivery of web fonts for first-time visitors. Five font loading strategies that use HTTP/2 Server Push and/or Preload were implemented on replicas of the landing pages from five real-world websites. The font loading strategies were evaluated against each other, and against the non-optimized version of each landing page. All the evaluated font loading strategies in this degree project improved the time it took to deliver the first web font content to the user’s screen, resulting in a faster FMP. It was also discovered that HTTP/2 Server Push, on its own, is not a more performance efficient resource prioritization approach than HTTP/2 Preload when it comes to delivering web font content to the client. Further, HTTP/2 Server Push and HTTP/2 Preload appears to be more efficient when used together, in the context of optimizing the delivery of web font content. However, all conclusions in this paper are based on the results gathered from testing the font loading strategies in an emulated environment and are yet to be confirmed on actual mobile devices with real network conditions.

Mount Meager est un complexe volcanique glaciaire en British Columbia (Canada). Il est connu pour ses glissements de terrain, dont celui de 2010 étant le plus grand glissement de terrain historique au Canada. Dans cette thèse, nous avons étudié les processus d'instabilités du volcan Mont Meager ainsi que les effets de la déglaciation en cours. Nous avons utilisé une approche pluridisciplinaire, intégrant la cartographie géologique, géomorphologique et structurelle, du terrain et de la télédétection, pour caractériser l'activité glaciaire et les glissements de terrain au Mount Meager. Nous avons utilisé la photogrammétrie Structure from Motion (SfM) et la technologie Lidar pour produire des modèles numériques de terrain, et techniques InSAR pour surveiller le mouvement et la déformation des pentes du volcan. Nous avons appliqué la technique SfM à des photographies aériennes historiques pour documenter les activités des glaciers et des glissements de terrain au Mount Meager. Nous avons discuté un modèle de croissance et d'érosion d'un volcan en période glaciaire et interglaciaire, ainsi que la valeur scientifique et de vulgarisation de la reconstruction topographique 3D. Nous avons décrit les dépôts de glissement de terrain de 2010 à Mount Meager pour interpréter la dynamique de leur mise en place. Le glissement de terrain de 2010 s'est divisé en phases riches en eau et pauvres en eau, ayant des distances d'écoulement différentes et des dépôts distincts. Nous avons analysé des photographies aériennes historiques remontant à 1948, afin de documenter la déformation de la pente avant l'effondrement de 2010. Le glacier situé a proximité du pied de la pente a reculé durant les années précédents la rupture. Cette effondrement a évolué en quatre sous-effondrements, impliquant toute la séquence volcanique et le socle. Nous avons estimé 6 × 106 m3 d'eau dans la pente, ce qui a permis la séparation de la phase frontale riche en eau. Le volume total d'effondrement est 53 ± 3.8 × 106 m3. Nous avons identifié 27 grands (>5×105 m2) flancs instables au Mount Meager et calculé a ~1.3 km3 de récession des glaciers depuis 1987. Le flanc ouest de Plinth Peak et de la vallée de Devastation Creek se sont déplacés de -34±10 mm -36±10 mm, respectivement, dans un période de 24 jours pendant l'été 2016. L’effondrement de ces flancs pourrait avoir un impact important sur les infrastructures et les communautés en aval du volcan. La décompression résultant de l'édifice volcanique après l'effondrement du flanc ouest de Plinth Peak affecterait le champ de contrainte à une profondeur de 6 km et jusqu'à 4 MPa. Cette décompression soudaine pourrait mener des éruptions hydrothermales et magmatiques. Un important glissement de terrain pourrait donc avoir joué un rôle dans le déclenchement de l'éruption de 2360 cal BP. / Mount Meager is a glacier-clad volcanic complex in British Columbia, Canada. It is known for its landslides, of which the 2010 is the largest Canadian historical landslide. In this thesis we investigated slope instability processes at Mount Meager volcano and the effects of ongoing deglaciation. We used a variety of methods including field and remote, geological, geomorphological and structural mapping to characterize glacial and landslide activity at Mount Meager. We used Structure from Motion photogrammetry (SfM) and Lidar to produce digital surface models and InSAR to monitor slope deformation. We applied SfM to historic photography to document glacier and landslide activity at Mount Meager. We discussed a model of growth and erosion of a volcano in glacial and interglacial periods, and the scientific and dissemination value of historic 3D topographic reconstruction. We described the 2010 Mount Meager landslide deposit to interpret emplacement dynamics and kinematics. The 2010 landslide separated in water-rich and water-poor phases that had different runout and distinct deposits. We analyzed historic airphotos to constrain the slope deformation prior to the 2010 collapse. The glacier near the toe of the slope retreated in the failure lead up, the collapse evolved in four subfailures involving the whole volcanic sequence and some basement rocks. We estimated 6 × 106 m3 of water in the slope, that allowed the separation of the frontal water-rich phase. The total failure volume was 53 ± 3.8 × 106 m3. We identified 27 large (>5×105 m2) unstable slopes at Mount Meager and calculated ~1.3 km3 of ice loss since 1987. The west flank of Plinth peak and Devastation Creek valley moved up to -34±10 mm and -36±10 mm, respectively, over a 24-day period during the summer of 2016. The failure of these slopes could impact infrastructures and communities downstream of the volcano. The resulting decompression on the volcanic edifice after the failure of Plinth peak would affect the stress field to a depth of 6 km and up to 4 MPa. This sudden decompression could lead to hydrothermal or magmatic eruptions.

The theoretical part focuses on the theme of the book - letter print - from the technical point of view. Further it explores the history, origin and various forms of a book. The thesis includes also chapters about a copyright book and the term " bibliophile". It also gives brief information about the graphic art, especially the technique of gravure. In the last part it aims at two personalities of Czech culture - a painter and graphic artist Jiří John and Jan Skácel. The practical part of the thesis is conceived as a folder that includes series of graphics accompanied with poetic texts. The graphic works are made by dry needle. The texts are written by Jan Skácel. The topic of the the copyright book expresses the author´s memories of her grandmother. The topics of each graphic work depicts the psychology of the contrast between the oldness and the youth and at the same time of the loss of a close person. The practical part should be also a sort of bibliophile.

Les grottes ornées paléolithiques de Combarelles, Font-de-Gaume, Bernifal et Rouffignac appartiennent au bassin versant de la Vézère, et se trouvent dans un rayon de quelques kilomètres autour du village des Eyzies (24, France). Leurs découvertes anciennes (de 1901 à 1956) sont à l'origine de nombreuses études ciblées qui ont souvent mis en évidence telle ou telle similitude, sans pour autant proposer une vision plus globale de ces sites majeurs de l'art magdalénien ouest-européen. Cette étude essaie donc de combler cette lacune et s'articule autour de 2 axes complémentaires de recherche. 1°- La méconnaissance de l'âge des œuvres et leur position chronologique relative nécessitent une révision des données issues du contexte archéologique des cavités. Cette approche conduit à mieux cerner les différentes fréquentations animales et humaines depuis le Pléistocène. Mais les résultats d'un site à l'autre demeurent contrastés. A Rouffignac et Bernifal, les vestiges attribuables au Paléolithique supérieur sont rares alors qu'ils sont plus nombreux aux Combarelles, mais restent difficiles à corréler avec l'art pariétal. A Font-de-Gaume, leur disparité compliquent les interprétations. Pour ce site, la découverte d'une exploitation de l'Ours (Ursus spaeleus) par l'Homme est un résultat notable. 2°- L'art de ces sites offre des points communs thématiques et stylistiques (notamment des figurations de mammouths et des signes tectiformes) qui justifient une étude comparative. La création d'un référentiel archéologique à partir de panneaux dont l'analyse externe garantit l'unité de composition, permet de découvrir que certaines variables morphologiques des figurations de mammouths et de bisons, sont plus aptes que d'autres à mettre en lumière le degré de parenté entre les dispositifs pariétaux. Une étude statistique des caractères morphologiques des figurations et l'analyse du bestiaire figuré permettent de compléter l'argumentation.

文字辨識為常見的電腦視覺應用之一，隨著正確率逐漸的上升，許多新的服務相繼出現，本論文改善了筆記管理軟體最主要的問題－文字切割，並提出兩種新的中文印刷體及手寫體的分類方法。我們將筆記文件中較常見的重點標記過濾後，再使用新核心的文字結構濾波取得筆記文件中的文字區塊，新的核心數據大幅降低原始核心的計算時間。本論文也使用文字結構濾波作為分辨印刷體、手寫體的特徵值，由於文字結構濾波會依據筆畫結構給予能量回饋，使得較工整的印刷體與手寫體能有所區別，此外也使用Sobel搭配不同角度範圍進行字體辨識，實驗結果證實了本論文所提出的文字切割及字體分類方法對於筆記文件資訊的處理是有效的。 / Character recognition is an important and practical application of computer vision. With the advance of this technology, more and more services embedding text recognition functionality have become available. However, segmentation is still the central issue in many situations. In this thesis, we tackle the character segmentation problem in note taking and management applications. We propose novel methods for the discrimination of handwritten and machine-printed Chinese characters. First, we perform noise removal using heuristics and apply a stroke filter with modified kernels to efficiently compute the bounding box for the text area. The responses of the stroke filter also serve as clues for differentiating machine-printed and handwritten texts. They are further enhanced using a SVM-based classifier that employs aggregated directional responses of edge detectors as input. Experiment results have validated the efficacy of the proposed approaches in terms of text localization and style recognition.