Surface Reflectance Recognition and Real-World Illumination Statistics

Humans distinguish materials such as metal, plastic, and paper effortlessly at a glance. Traditional computer vision systems cannot solve this problem at all. Recognizing surface reflectance properties from a single photograph is difficult because the observed image depends heavily on the amount of light incident from every direction. A mirrored sphere, for example, produces a different image in every environment. To make matters worse, two surfaces with different reflectance properties could produce identical images. The mirrored sphere simply reflects its surroundings, so in the right artificial setting, it could mimic the appearance of a matte ping-pong ball. Yet, humans possess an intuitive sense of what materials typically "look like" in the real world. This thesis develops computational algorithms with a similar ability to recognize reflectance properties from photographs under unknown, real-world illumination conditions. Real-world illumination is complex, with light typically incident on a surface from every direction. We find, however, that real-world illumination patterns are not arbitrary. They exhibit highly predictable spatial structure, which we describe largely in the wavelet domain. Although they differ in several respects from the typical photographs, illumination patterns share much of the regularity described in the natural image statistics literature. These properties of real-world illumination lead to predictable image statistics for a surface with given reflectance properties. We construct a system that classifies a surface according to its reflectance from a single photograph under unknown illuminination. Our algorithm learns relationships between surface reflectance and certain statistics computed from the observed image. Like the human visual system, we solve the otherwise underconstrained inverse problem of reflectance estimation by taking advantage of the statistical regularity of illumination. For surfaces with homogeneous reflectance properties and known geometry, our system rivals human performance.

Computer-aided diagnosis of carotid atherosclerosis based on ultrasound image statistics, laws' texture and neural networks

Quantitative characterisation of carotid atherosclerosis and classification into symptomatic or asymptomatic is crucial in planning optimal treatment of atheromatous plaque. The computer-aided diagnosis (CAD) system described in this paper can analyse ultrasound (US) images of carotid artery and classify them into symptomatic or asymptomatic based on their echogenicity characteristics. The CAD system consists of three modules: a) the feature extraction module, where first-order statistical (FOS) features and Laws' texture energy can be estimated, b) the dimensionality reduction module, where the number of features can be reduced using analysis of variance (ANOVA), and c) the classifier module consisting of a neural network (NN) trained by a novel hybrid method based on genetic algorithms (GAs) along with the back propagation algorithm. The hybrid method is able to select the most robust features, to adjust automatically the NN architecture and to optimise the classification performance. The performance is measured by the accuracy, sensitivity, specificity and the area under the receiver-operating characteristic (ROC) curve. The CAD design and development is based on images from 54 symptomatic and 54 asymptomatic plaques. This study demonstrates the ability of a CAD system based on US image analysis and a hybrid trained NN to identify atheromatous plaques at high risk of stroke.

A Logarithmic Image Prior for Blind Deconvolution

Blind Deconvolution consists in the estimation of a sharp image and a blur kernel from an observed blurry image. Because the blur model admits several solutions it is necessary to devise an image prior that favors the true blur kernel and sharp image. Many successful image priors enforce the sparsity of the sharp image gradients. Ideally the L0 “norm” is the best choice for promoting sparsity, but because it is computationally intractable, some methods have used a logarithmic approximation. In this work we also study a logarithmic image prior. We show empirically how well the prior suits the blind deconvolution problem. Our analysis confirms experimentally the hypothesis that a prior should not necessarily model natural image statistics to correctly estimate the blur kernel. Furthermore, we show that a simple Maximum a Posteriori formulation is enough to achieve state of the art results. To minimize such formulation we devise two iterative minimization algorithms that cope with the non-convexity of the logarithmic prior: one obtained via the primal-dual approach and one via majorization-minimization.

The standard dictionary of facts; history, language, literature, biography, geography, travel, art, government, politics, industry, invention, commerce, science, education, natural history, statistics and miscellany,

Mode of access: Internet.

A history of Kentucky, embracing gleanings, reminiscences, antiquities, natural curiosities, statistics, and biographical sketches.

Microfilm. Ann Arbor, Mich., University Microfilms [n.d.] (American culture series, Reel 425.14)

The standard dictionary of facts; history, language, literature, biography, geography, travel, art, government, politics, industry, invention, commerce, science, education, natural history, statistics, miscellany; a practical handbook of ready reference based upon everyday needs,

Mode of access: Internet.

The standard dictionary of facts;, history, language, literature, biography, geography, travel, art, government, politics, industry, invention, commerce, science, education, natural history, statistics and miscellany; a practical handbook of ready reference based upon everyday needs,

Mode of access: Internet.

Human sensitivity to phase perturbations in natural images:A statistical framework

Fourier-phase information is important in determining the appearance of natural scenes, but the structure of natural-image phase spectra is highly complex and difficult to relate directly to human perceptual processes. This problem is addressed by extending previous investigations of human visual sensitivity to the randomisation and quantisation of Fourier phase in natural images. The salience of the image changes induced by these physical processes is shown to depend critically on the nature of the original phase spectrum of each image, and the processes of randomisation and quantisation are shown to be perceptually equivalent provided that they shift image phase components by the same average amount. These results are explained by assuming that the visual system is sensitive to those phase-domain image changes which also alter certain global higher-order image statistics. This assumption may be used to place constraints on the likely nature of cortical processing: mechanisms which correlate the outputs of a bank of relative-phase-sensitive units are found to be consistent with the patterns of sensitivity reported here.

No-reference image quality assessment in contourlet domain

The target of no-reference (NR) image quality assessment (IQA) is to establish a computational model to predict the visual quality of an image. The existing prominent method is based on natural scene statistics (NSS). It uses the joint and marginal distributions of wavelet coefficients for IQA. However, this method is only applicable to JPEG2000 compressed images. Since the wavelet transform fails to capture the directional information of images, an improved NSS model is established by contourlets. In this paper, the contourlet transform is utilized to NSS of images, and then the relationship of contourlet coefficients is represented by the joint distribution. The statistics of contourlet coefficients are applicable to indicate variation of image quality. In addition, an image-dependent threshold is adopted to reduce the effect of content to the statistical model. Finally, image quality can be evaluated by combining the extracted features in each subband nonlinearly. Our algorithm is trained and tested on the LIVE database II. Experimental results demonstrate that the proposed algorithm is superior to the conventional NSS model and can be applied to different distortions. © 2009 Elsevier B.V. All rights reserved.

Perception of global image contrast is predicted by the same spatial integration model of gain control as detection and discrimination

How are the image statistics of global image contrast computed? We answered this by using a contrast-matching task for checkerboard configurations of ‘battenberg’ micro-patterns where the contrasts and spatial spreads of interdigitated pairs of micro-patterns were adjusted independently. Test stimuli were 20 × 20 arrays with various sized cluster widths, matched to standard patterns of uniform contrast. When one of the test patterns contained a pattern with much higher contrast than the other, that determined global pattern contrast, as in a max() operation. Crucially, however, the full matching functions had a curious intermediate region where low contrast additions for one pattern to intermediate contrasts of the other caused a paradoxical reduction in perceived global contrast. None of the following models predicted this: RMS, energy, linear sum, max, Legge and Foley. However, a gain control model incorporating wide-field integration and suppression of nonlinear contrast responses predicted the results with no free parameters. This model was derived from experiments on summation of contrast at threshold, and masking and summation effects in dipper functions. Those experiments were also inconsistent with the failed models above. Thus, we conclude that our contrast gain control model (Meese & Summers, 2007) describes a fundamental operation in human contrast vision.

Variational and learning models for image and time series inverse problems

Inverse problems are at the core of many challenging applications. Variational and learning models provide estimated solutions of inverse problems as the outcome of specific reconstruction maps. In the variational approach, the result of the reconstruction map is the solution of a regularized minimization problem encoding information on the acquisition process and prior knowledge on the solution. In the learning approach, the reconstruction map is a parametric function whose parameters are identified by solving a minimization problem depending on a large set of data. In this thesis, we go beyond this apparent dichotomy between variational and learning models and we show they can be harmoniously merged in unified hybrid frameworks preserving their main advantages. We develop several highly efficient methods based on both these model-driven and data-driven strategies, for which we provide a detailed convergence analysis. The arising algorithms are applied to solve inverse problems involving images and time series. For each task, we show the proposed schemes improve the performances of many other existing methods in terms of both computational burden and quality of the solution. In the first part, we focus on gradient-based regularized variational models which are shown to be effective for segmentation purposes and thermal and medical image enhancement. We consider gradient sparsity-promoting regularized models for which we develop different strategies to estimate the regularization strength. Furthermore, we introduce a novel gradient-based Plug-and-Play convergent scheme considering a deep learning based denoiser trained on the gradient domain. In the second part, we address the tasks of natural image deblurring, image and video super resolution microscopy and positioning time series prediction, through deep learning based methods. We boost the performances of supervised, such as trained convolutional and recurrent networks, and unsupervised deep learning strategies, such as Deep Image Prior, by penalizing the losses with handcrafted regularization terms.

A bio-inspired computational model for motion detection

Tese de Doutoramento (Programa Doutoral em Engenharia Biomédica)

Density, climate and varying return points: an analysis of long-term population fluctuations in the threatened European tree frog.

Experimental research has identified many putative agents of amphibian decline, yet the population-level consequences of these agents remain unknown, owing to lack of information on compensatory density dependence in natural populations. Here, we investigate the relative importance of intrinsic (density-dependent) and extrinsic (climatic) factors impacting the dynamics of a tree frog (Hyla arborea) population over 22 years. A combination of log-linear density dependence and rainfall (with a 2-year time lag corresponding to development time) explain 75% of the variance in the rate of increase. Such fluctuations around a variable return point might be responsible for the seemingly erratic demography and disequilibrium dynamics of many amphibian populations.

On some aspects of coherent risk measures and their applications

Le sujet principal de cette thèse porte sur les mesures de risque. L'objectif général est d'investiguer certains aspects des mesures de risque dans les applications financières. Le cadre théorique de ce travail est celui des mesures cohérentes de risque telle que définie dans Artzner et al (1999). Mais ce n'est pas la seule classe de mesure du risque que nous étudions. Par exemple, nous étudions aussi quelques aspects des "statistiques naturelles de risque" (en anglais natural risk statistics) Kou et al (2006) et des mesures convexes du risque Follmer and Schied(2002). Les contributions principales de cette thèse peuvent être regroupées selon trois axes: allocation de capital, évaluation des risques et capital requis et solvabilité. Dans le chapitre 2 nous caractérisons les mesures de risque avec la propriété de Lebesgue sur l'ensemble des processus bornés càdlàg (continu à droite, limité à gauche). Cette caractérisation nous permet de présenter deux applications dans l'évaluation des risques et l'allocation de capital. Dans le chapitre 3, nous étendons la notion de statistiques naturelles de risque à l'espace des suites infinies. Cette généralisation nous permet de construire de façon cohérente des mesures de risque pour des bases de données de n'importe quelle taille. Dans le chapitre 4, nous discutons le concept de "bonnes affaires" (en anglais Good Deals), pour notamment caractériser les situations du marché où ces positions pathologiques sont présentes. Finalement, dans le chapitre 5, nous essayons de relier les trois chapitres en étendant la définition de "bonnes affaires" dans un cadre plus large qui comprendrait les mesures de risque analysées dans les chapitres 2 et 3.