Scene Classification Using a Hybrid Generative/Discriminative Approach

We investigate whether dimensionality reduction using a latent generative model is beneficial for the task of weakly supervised scene classification. In detail, we are given a set of labeled images of scenes (for example, coast, forest, city, river, etc.), and our objective is to classify a new image into one of these categories. Our approach consists of first discovering latent ";topics"; using probabilistic Latent Semantic Analysis (pLSA), a generative model from the statistical text literature here applied to a bag of visual words representation for each image, and subsequently, training a multiway classifier on the topic distribution vector for each image. We compare this approach to that of representing each image by a bag of visual words vector directly and training a multiway classifier on these vectors. To this end, we introduce a novel vocabulary using dense color SIFT descriptors and then investigate the classification performance under changes in the size of the visual vocabulary, the number of latent topics learned, and the type of discriminative classifier used (k-nearest neighbor or SVM). We achieve superior classification performance to recent publications that have used a bag of visual word representation, in all cases, using the authors' own data sets and testing protocols. We also investigate the gain in adding spatial information. We show applications to image retrieval with relevance feedback and to scene classification in videos

Recovering Euclidean deformable models from stereo-motion

In this paper we present a novel structure from motion (SfM) approach able to infer 3D deformable models from uncalibrated stereo images. Using a stereo setup dramatically improves the 3D model estimation when the observed 3D shape is mostly deforming without undergoing strong rigid motion. Our approach first calibrates the stereo system automatically and then computes a single metric rigid structure for each frame. Afterwards, these 3D shapes are aligned to a reference view using a RANSAC method in order to compute the mean shape of the object and to select the subset of points on the object which have remained rigid throughout the sequence without deforming. The selected rigid points are then used to compute frame-wise shape registration and to extract the motion parameters robustly from frame to frame. Finally, all this information is used in a global optimization stage with bundle adjustment which allows to refine the frame-wise initial solution and also to recover the non-rigid 3D model. We show results on synthetic and real data that prove the performance of the proposed method even when there is no rigid motion in the original sequence

Model-based objects recognition in industrial environments for autonomous vehicles control

Behavior-based navigation of autonomous vehicles requires the recognition of the navigable areas and the potential obstacles. In this paper we describe a model-based objects recognition system which is part of an image interpretation system intended to assist the navigation of autonomous vehicles that operate in industrial environments. The recognition system integrates color, shape and texture information together with the location of the vanishing point. The recognition process starts from some prior scene knowledge, that is, a generic model of the expected scene and the potential objects. The recognition system constitutes an approach where different low-level vision techniques extract a multitude of image descriptors which are then analyzed using a rule-based reasoning system to interpret the image content. This system has been implemented using a rule-based cooperative expert system

Model-based objects recognition in man-made environments

We describe a model-based objects recognition system which is part of an image interpretation system intended to assist autonomous vehicles navigation. The system is intended to operate in man-made environments. Behavior-based navigation of autonomous vehicles involves the recognition of navigable areas and the potential obstacles. The recognition system integrates color, shape and texture information together with the location of the vanishing point. The recognition process starts from some prior scene knowledge, that is, a generic model of the expected scene and the potential objects. The recognition system constitutes an approach where different low-level vision techniques extract a multitude of image descriptors which are then analyzed using a rule-based reasoning system to interpret the image content. This system has been implemented using CEES, the C++ embedded expert system shell developed in the Systems Engineering and Automatic Control Laboratory (University of Girona) as a specific rule-based problem solving tool. It has been especially conceived for supporting cooperative expert systems, and uses the object oriented programming paradigm

A new optimised De Bruijn coding strategy for structured light patterns

Coded structured light is an optical technique based on active stereovision that obtains the shape of objects. One shot techniques are based on projecting a unique light pattern with an LCD projector so that grabbing an image with a camera, a large number of correspondences can be obtained. Then, a 3D reconstruction of the illuminated object can be recovered by means of triangulation. The most used strategy to encode one-shot patterns is based on De Bruijn sequences. In This work a new way to design patterns using this type of sequences is presented. The new coding strategy minimises the number of required colours and maximises both the resolution and the accuracy

Use of fused airborne scanning laser altimetry and digital map data for urban flood modelling

Flood modelling of urban areas is still at an early stage, partly because until recently topographic data of sufficiently high resolution and accuracy have been lacking in urban areas. However, Digital Surface Models (DSMs) generated from airborne scanning laser altimetry (LiDAR) having sub-metre spatial resolution have now become available, and these are able to represent the complexities of urban topography. The paper describes the development of a LiDAR post-processor for urban flood modelling based on the fusion of LiDAR and digital map data. The map data are used in conjunction with LiDAR data to identify different object types in urban areas, though pattern recognition techniques are also employed. Post-processing produces a Digital Terrain Model (DTM) for use as model bathymetry, and also a friction parameter map for use in estimating spatially-distributed friction coefficients. In vegetated areas, friction is estimated from LiDAR-derived vegetation height, and (unlike most vegetation removal software) the method copes with short vegetation less than ~1m high, which may occupy a substantial fraction of even an urban floodplain. The DTM and friction parameter map may also be used to help to generate an unstructured mesh of a vegetated urban floodplain for use by a 2D finite element model. The mesh is decomposed to reflect floodplain features having different frictional properties to their surroundings, including urban features such as buildings and roads as well as taller vegetation features such as trees and hedges. This allows a more accurate estimation of local friction. The method produces a substantial node density due to the small dimensions of many urban features.

Resource allocation analysis of the stochastic diffusion search

The Stochastic Diffusion Search (SDS) was developed as a solution to the best-fit search problem. Thus, as a special case it is capable of solving the transform invariant pattern recognition problem. SDS is efficient and, although inherently probabilistic, produces very reliable solutions in widely ranging search conditions. However, to date a systematic formal investigation of its properties has not been carried out. This thesis addresses this problem. The thesis reports results pertaining to the global convergence of SDS as well as characterising its time complexity. However, the main emphasis of the work, reports on the resource allocation aspect of the Stochastic Diffusion Search operations. The thesis introduces a novel model of the algorithm, generalising an Ehrenfest Urn Model from statistical physics. This approach makes it possible to obtain a thorough characterisation of the response of the algorithm in terms of the parameters describing the search conditions in case of a unique best-fit pattern in the search space. This model is further generalised in order to account for different search conditions: two solutions in the search space and search for a unique solution in a noisy search space. Also an approximate solution in the case of two alternative solutions is proposed and compared with predictions of the extended Ehrenfest Urn model. The analysis performed enabled a quantitative characterisation of the Stochastic Diffusion Search in terms of exploration and exploitation of the search space. It appeared that SDS is biased towards the latter mode of operation. This novel perspective on the Stochastic Diffusion Search lead to an investigation of extensions of the standard SDS, which would strike a different balance between these two modes of search space processing. Thus, two novel algorithms were derived from the standard Stochastic Diffusion Search, ‘context-free’ and ‘context-sensitive’ SDS, and their properties were analysed with respect to resource allocation. It appeared that they shared some of the desired features of their predecessor but also possessed some properties not present in the classic SDS. The theory developed in the thesis was illustrated throughout with carefully chosen simulations of a best-fit search for a string pattern, a simple but representative domain, enabling careful control of search conditions.

Type-1 error inflation in the traditional by-participant analysis to metamemory accuracy: a generalized mixed-effects model perspective

In order to examine metacognitive accuracy (i.e., the relationship between metacognitive judgment and memory performance), researchers often rely on by-participant analysis, where metacognitive accuracy (e.g., resolution, as measured by the gamma coefficient or signal detection measures) is computed for each participant and the computed values are entered into group-level statistical tests such as the t-test. In the current work, we argue that the by-participant analysis, regardless of the accuracy measurements used, would produce a substantial inflation of Type-1 error rates, when a random item effect is present. A mixed-effects model is proposed as a way to effectively address the issue, and our simulation studies examining Type-1 error rates indeed showed superior performance of mixed-effects model analysis as compared to the conventional by-participant analysis. We also present real data applications to illustrate further strengths of mixed-effects model analysis. Our findings imply that caution is needed when using the by-participant analysis, and recommend the mixed-effects model analysis.

A complex network-based approach for boundary shape analysis

This paper introduces a novel methodology to shape boundary characterization, where a shape is modeled into a small-world complex network. It uses degree and joint degree measurements in a dynamic evolution network to compose a set of shape descriptors. The proposed shape characterization method has all efficient power of shape characterization, it is robust, noise tolerant, scale invariant and rotation invariant. A leaf plant classification experiment is presented on three image databases in order to evaluate the method and compare it with other descriptors in the literature (Fourier descriptors, Curvature, Zernike moments and multiscale fractal dimension). (C) 2008 Elsevier Ltd. All rights reserved.

Multiscale Analysis of Topographic Surface Roughness in the Midland Valley, Scotland

Surface roughness is an important geomorphological variable which has been used in the Earth and planetary sciences to infer material properties, current/past processes, and the time elapsed since formation. No single definition exists; however, within the context of geomorphometry, we use surface roughness as an expression of the variability of a topographic surface at a given scale, where the scale of analysis is determined by the size of the landforms or geomorphic features of interest. Six techniques for the calculation of surface roughness were selected for an assessment of the parameter`s behavior at different spatial scales and data-set resolutions. Area ratio operated independently of scale, providing consistent results across spatial resolutions. Vector dispersion produced results with increasing roughness and homogenization of terrain at coarser resolutions and larger window sizes. Standard deviation of residual topography highlighted local features and did not detect regional relief. Standard deviation of elevation correctly identified breaks of slope and was good at detecting regional relief. Standard deviation of slope (SD(slope)) also correctly identified smooth sloping areas and breaks of slope, providing the best results for geomorphological analysis. Standard deviation of profile curvature identified the breaks of slope, although not as strongly as SD(slope), and it is sensitive to noise and spurious data. In general, SD(slope) offered good performance at a variety of scales, while the simplicity of calculation is perhaps its single greatest benefit.

Análise da rede de atores na trajetória de implantação de lan houses no Brasil e seu possível papel de agente propiciador de inclusão digital

A Lan House, surgida no Brasil como um meio de entretenimento para os jovens, se tornou, em um curto espaço de tempo, uma febre nas periferias das grandes cidades brasileiras. Essa disseminação se deu, principalmente, após o programa “Computador para Todos” lançado pelo Governo Federal a título de política pública de inclusão digital. Assim, as Lan Houses se constituíram em uma oportunidade de acesso ao computador e à Internet para aqueles que não teriam ingresso à rede se não fosse a existência desse tipo de instituição comercial (CDI, 2010), sendo a segunda principal provedora de acesso público às TIC no país (CETIC, 2010). Diante desse cenário, este estudo se propõe a descrever a trajetória na implantação das Lan Houses no Brasil, sob a ótica da Teoria Ator-Rede, identificando os atores relevantes na formação de uma rede sociotécnica, por meio do método de estudo de caso único realizado no bairro Jardim Catarina, em São Gonçalo. O trabalho apresenta, ainda, o modelo heurístico de inclusão digital para avaliar se este tipo de estabelecimento apresenta fatores relevantes para fomentar a inclusão digital. O resultado desta análise revela que as Lan Houses não são um agente de inclusão digital, apesar de sua relevância para as regiões com menores índices de renda e, por conseguinte, restritas ao uso de computadores e Internet, como o bairro Jardim Catarina.

Estado nutricional em nitrogênio da grama esmeralda avaliado por meio do teor foliar, clorofilômetro e imagem digital, em área adubada com lodo de esgoto

A intensidade da cor verde da folha pode ser alternativa para estimar a concentração de N na planta, devido à relação entre o teor de clorofila e o de N no tecido foliar. Objetivou-se neste trabalho avaliar índices da cor verde da grama esmeralda obtidos da análise da imagem digital e pelo uso do clorofilômetro para predizer o estado nutricional em N fornecido pelo lodo de esgoto. O experimento foi instalado e desenvolvido em uma propriedade comercial de grama esmeralda, localizada na cidade de Itapetininga (SP). O delineamento experimental foi em blocos casualizados, com quatro repetições e cinco doses de lodo de esgoto: 0, 10, 20, 30 e 40 Mg ha-1, base seca. As doses de lodo aplicadas correspondem a 100, 200, 300 e 400 kg ha-1 de nitrogênio disponível. Foram avaliadas as concentrações de N e a intensidade de coloração verde da folha pelo uso do clorofilômetro (ICV) e por meio da análise da imagem digital (G, H e ICVE) aos 45, 105 e 165 dias após a aplicação do lodo. Os valores de intensidade obtidos foram correlacionados com a concentração de N na lâmina foliar e com a taxa de cobertura do solo determinada nas mesmas épocas. A aplicação de doses de lodo de esgoto proporcionou aumento dos índices de cor verde e da concentração de N nas folhas da grama esmeralda. A concentração de N na lâmina foliar pode auxiliar a adubação nitrogenada em cobertura, pois proporcionou altas correlações com a taxa de cobertura do solo. O matiz (H) obtido com a imagem digital e a intensidade de cor verde da folha (ICV) obtida com o clorofilômetro correlacionaram-se com a concentração de N e com a taxa de cobertura do solo e, dessa forma, podem servir como índices na recomendação da adubação nitrogenada.

A low-cost 3D reconstruction system using a single-shot projection of a pattern matrix

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Úlceras crônicas dos membros inferiores: avaliação pela fotografia digital

OBJETIVO: Avaliar o desempenho da análise de imagem digital na estimativa da área acometida pelas úlceras crônicas dos membros inferiores. MÉTODOS: Estudo prospectivo em que foram mensuradas úlceras empregando o método planimétrico clássico, utilizando desenho dos seus contornos em filme plástico transparente, medida sua área posteriormente por folha milimetrada. Esses valores foram utilizados como padrão para a comparação com a estimativa de área pelas fotografias digitais padronizadas das úlceras e dos desenhos das mesmas em filme plástico. Para criar um referencial de conversão dos pixels em milímetros, foi empregado um adesivo com tamanho conhecido, adjacente à úlcera. RESULTADOS: foram avaliadas 42 lesões em 20 pacientes portadores de úlceras crônicas de membros inferiores. As áreas das úlceras variaram de 0,24 a 101,65cm². Observou-se forte correlação entre as medidas planimétricas e as fotos das úlceras (R²=0,86 p<0,01), porém a correlação das medidas planimétricas com as fotos digitais dos desenhos das úlceras foi ainda maior (R²=0,99 p<0,01). CONCLUSÃO: A fotografia digital padronizada revelou-se método rápido, preciso e não-invasivo capaz de estimar a área afetada por úlceras. A avaliação das medidas fotográficas dos contornos das úlceras deve ser preferida em relação à análise de sua fotografia direta.

Imageamento digital de paleocavernas colapsadas com georadar

In this study, the methodological procedures involved in digital imaging of collapsed paleocaves in tufa using GPR are presented. These carbonate deposits occur in the Quixeré region, Ceará State (NE Brazil), on the western border of the Potiguar Basin. Collapsed paleocaves are exposed along a state road, which were selected to this study. We chose a portion of the called Quixeré outcrop for making a photomosaic and caring out a GPR test section to compare and parameterize the karst geometries on the geophysical line. The results were satisfactory and led to the adoption of criteria for the interpretation of others GPR sections acquired in the region of the Quixeré outcrop. Two grids of GPR lines were acquired; the first one was wider and more spaced and guided the location of the second grid, denser and located in the southern part of the outcrop. The radargrams of the second grid reveal satisfactorily the collapsed paleocaves geometries. For each grid has been developed a digital solid model of the Quixeré outcrop. The first model allows the recognition of the general distribution and location of collapsed paleocaves in tufa deposits, while the second more detailed digital model provides not only the 3D individualization of the major paleocaves, but also the estimation of their respective volumes. The digital solid models are presented here as a new frontier in the study of analog outcrops to reservoirs (for groundwater and hydrocarbon), in which the volumetric parameterization and characterization of geological bodies become essential for composing the databases, which together with petrophysical properties information, are used in more realistic computer simulations for sedimentary reservoirs.