No description available.

The statistical structure of natural images has sparked interest in regards to both visual processing of such images and the aesthetic responses they elicit. We bring together these two lines of research under the simple proposition that early visual processing, more specifically visual sensitivity, may mediate visual preferences. In our first set of experiments, we measured both visual preference and visual sensitivity in the same observers, using patterns varying in the slope of their amplitude spectrum. Our results found similar functions for both variables, supporting our hypothesis. In a second set of experiments, we investigated the same question with regards to simple spatial frequency variations, and found even more convincing results. We conclude that early visual processing is implicated in visual preference, supporting the recently emerging sensory-based approaches to understanding visual preference.

<p>Predmet istraživanja ove doktorske disertacije je problem<br />interpolacije slike. Glavni fokus disertacije je interpolacija slike<br />uz očuvanje prirodnosti teksture i očuvanje ivica (oštrine)<br />interpolirane slike. Dodatni izazov je da algoritam za interpolaciju<br />slike bude pogodan za primenu u uređajima sa ograničenim resursima.<br />Kvalitet rešenja se ocenjuje poređenjem sa algoritmima poznatim u<br />dostupnoj literaturi korišćenjem odgovarajućih metrika.</p> / <p>This PhD dissertation addresses the problem of image interpolation. The main<br />focus of the dissertation is image interpolation algorithm which preserves<br />edges and keeps a natural texture of interpolated images. Additional challenge<br />for image interpolation algorithm is to be suitable for application on resourcelimited<br />platforms. The quality of the proposed solution is benchmarked against<br />known image interpolation algorithms using appropriate metrics.</p>

Dans cette thèse, nous nous intéressons aux structures d’une image en général, et plus particulièrement aux gradients et aux contours. Ces dernières se sont révélées très importantes ces dernières années pour de nombreuses applications en infographie,telles que la réalité augmentée et la stylisation d’images et de vidéos. Le but de toute analyse des structures d’une image est de décrire à un haut-niveau la compréhension que l’on peut avoir de son contenu et de fournir les bases nécessaires à l’amélioration de la qualité des applications citées au-dessus, notamment la lisibilité, la précision, la cohérence spatiale et temporelle.Dans une premier temps, nous démontrons le rôle important que ces structures jouent pour des applications de type composition “Focus+Context”. Une telle approche est utilisée en réalité augmentée pour permettre la visualisation de parties d’une scènes qui sont normalement derrières ce que l’on peut observer dans un flux vidéo. L’utilisation d’une segmentation et de lignes caractéristiques permettent de mettre en avant et/ou de révéler les relations d’ordre entre les différents objets de la scène. Pour la synthèse d’images guidée par une fonction d’importance, de multiples styles de rendu sont combinés de manière cohérente grâce à l’utilisation d’une carte de gradients et une de saillance.Dans un deuxième temps, nous introduisons une nouvelle techniques qui permet de reconstruire de manière continue un champ de gradient, et ceci sans trop lisser les détails originaux contenus dans l’image. Pour cela, nous développons une nouvelle méthode d’approximation locale et de plus haut-degré pour des champs de gradients discrets et non-orientés. Cette méthode est basée sur le formalisme“moving least square” (MLS). Nous démontrons que notre approximation isotrope et linéaire est de meilleure qualité que le classique tenseur de structure : les détails sont mieux préservés et les instabilités sont réduites de manière significative. Nous démontrons aussi que notre nouveau champ de gradients apporte des améliorations à de nombreuses techniques de stylisation.Finalement, nous démontrons que l’utilisation d’une technique d’analyse de profil caractéristique par approximation polynomiale permet de distinguer les variations douces des zones dures. Les paramètres du profil sont utilisés comme des paramètres de stylisation tels que l’orientation des coups de pinceau, leur taille et leur opacité. Cela permet la création d’une large variété de styles de ligne. / In this thesis we consider in general image structures and more specifically, imagegradient and contours. They have been proven useful in recent years for variouscomputer graphics applications, such as Augmented Reality (AR), image and videostylization. The goal of analyzing image structures is to describe a high level understandingof image contents and to provide a powerful support to improve thequality of applications, such as visual legibility, accuracy, spatial and temporal coherence.We first demonstrate the important role of image structures in Focus+Contextcompositing. For Focus+Context rendering in AR, a technique dedicated to the visualizationof hidden scenes in video streams, the use of screen segmentation andfeature lines significantly emphasizes the depth cues of occluded scenes, and revealsthe correct occluding order. For importance-driven image synthesis, multiplerendering styles are combined in a coherent manner by using image gradient fieldand saliency map.In the second part, we thus introduce a new approach to estimate a continuousgradient field without oversmoothing the original details contained in an image.For this purpose, we develop a new and higher-order local approximation methodof discrete non-oriented gradient fields based on a moving least square (MLS) formalism.We show that our isotropic linear approximation outperforms classicalstructure tensor: image details are better preserved and instabilities are significantlyreduced. We demonstrate how our non-oriented MLS gradient field benefitsto various image stylization approaches.Finally, we demonstrate that the use of a feature profile analysis for image lineextraction via fitting techniques permits to distinguish sharp and smooth features.Profile parameters are then mapped to stylistic parameters such as brush orientation,size or opacity to give rise to a wide range of line-based styles.

La presente investigación tiene como objetivo analizar la representación del conflicto de los personajes protagónicos a través de la estructura de la imagen en la película animada japonesa Una voz Silenciosa (2016) dirigida por Naoko Yamada. En la película se presentan los niveles de conflictos del personaje para evidenciar en la narrativa los problemas sociales que son tratados como tema principal de la trama. De ese modo, los componentes visuales son una clave fundamental para lograr representar visualmente la construcción de los protagonistas Shoya y Shouko, fortaleciendo así su arco de transformación debido a los acontecimientos por los que atraviesan en la historia. / The present research aims to analyze the representation of the conflict of the protagonist characters through the image structure in the Japanese animated film A Silent Voice (2016) directed by Naoko Yamada. In the film, the levels of character conflicts are presented to evidence in the narrative the social problems that are treated as the main theme of the plot. Thus, the visual components are a fundamental key to visually represent the construction of the protagonists Shoya and Shouko, strengthening their arc of transformation due to the events they go through in the story. / Trabajo de investigación

Image segmentation is one of the most difficult tasks in image processing. Segmentation algorithms are generally based on searching a region where pixels share similar gray level intensity and satisfy a set of defined criteria. However, the segmented region cannot be used directly for partial image retrieval. In this dissertation, a Contour Based Image Structure (CBIS) model is introduced. In this model, images are divided into several objects defined by their bounding contours. The bounding contour structure allows individual object extraction, and partial object matching and retrieval from a standard CBIS image structure. The CBIS model allows the representation of 3D objects by their bounding contours which is suitable for parallel implementation particularly when extracting contour features and matching them for 3D images require heavy computations. This computational burden becomes worse for images with high resolution and large contour density. In this essence we designed two parallel algorithms; Contour Parallelization Algorithm (CPA) and Partial Retrieval Parallelization Algorithm (PRPA). Both algorithms have considerably improved the performance of CBIS for both contour shape matching as well as partial image retrieval. To improve the effectiveness of CBIS in segmenting images with inhomogeneous backgrounds we used the phase congruency invariant features of Fourier transform components to highlight boundaries of objects prior to extracting their contours. The contour matching process has also been improved by constructing a fuzzy contour matching system that allows unbiased matching decisions. Further improvements have been achieved through the use of a contour tailored Fourier descriptor to make translation and rotation invariance. It is proved to be suitable for general contour shape matching where translation, rotation, and scaling invariance are required. For those images which are hard to be classified by object contours such as bacterial images, we define a multi-level cosine transform to extract their texture features for image classification. The low frequency Discrete Cosine Transform coefficients and Zenike moments derived from images are trained by Support Vector Machine (SVM) to generate multiple classifiers.

This thesis, within the subfield of computer science known as computer vision, deals with the use of scale-space analysis in early low-level processing of visual information. The main contributions comprise the following five subjects: The formulation of a scale-space theory for discrete signals. Previously, the scale-space concept has been expressed for continuous signals only. We propose that the canonical way to construct a scale-space for discrete signals is by convolution with a kernel called the discrete analogue of the Gaussian kernel, or equivalently by solving a semi-discretized version of the diffusion equation. Both the one-dimensional and two-dimensional cases are covered. An extensive analysis of discrete smoothing kernels is carried out for one-dimensional signals and the discrete scale-space properties of the most common discretizations to the continuous theory are analysed. A representation, called the scale-space primal sketch, which gives a formal description of the hierarchical relations between structures at different levels of scale. It is aimed at making information in the scale-space representation explicit. We give a theory for its construction and an algorithm for computing it. A theory for extracting significant image structures and determining the scales of these structures from this representation in a solely bottom-up data-driven way. Examples demonstrating how such qualitative information extracted from the scale-space primal sketch can be used for guiding and simplifying other early visual processes. Applications are given to edge detection, histogram analysis and classification based on local features. Among other possible applications one can mention perceptual grouping, texture analysis, stereo matching, model matching and motion. A detailed theoretical analysis of the evolution properties of critical points and blobs in scale-space, comprising drift velocity estimates under scale-space smoothing, a classification of the possible types of generic events at bifurcation situations and estimates of how the number of local extrema in a signal can be expected to decrease as function of the scale parameter. For two-dimensional signals the generic bifurcation events are annihilations and creations of extremum-saddle point pairs. Interpreted in terms of blobs, these transitions correspond to annihilations, merges, splits and creations. Experiments on different types of real imagery demonstrate that the proposed theory gives perceptually intuitive results. / <p>QC 20120119</p>

For the purpose of this study, the geometric images refer to the results obtained by reproducing the percept of elements that constitute the content of geometry. The scientific problem addressed in this study is the achievement of more effective geometric images formation at primary school stages on the basis of holistic approach to the geometric education. The study objective is to get deeper knowledge of the mechanism on the formation of geometric images at primary school. The theoretical analysis includes the definition of geometric image conception worked out based on the works of scientists representing a variety of different disciplines, the overview of the existing situation with respect to the formation of geometric images in pupils of primary school and modelling of geometric images formation process at primary school stage based on the results of theoretical analysis. The experimental study on the development of geometric images at primary school stages was planned based on the hypothetical geometric images formation model. The study has been narrowed down to the evaluation of the impact that the manual constructive geometry has upon the geometric images of schoolchildren. The empirical reasoning of dissertation, in total: questionnaire data completed by 653 schoolchildren and qualitative data collected in the result of formative experiment. The results of the study may be applied in the process of training future teachers of primary school. They may also be... [to full text]