Higher-order regularization and morphological techniques for image segmentation

La segmentación de imágenes es un campo importante de la visión computacional y una de las áreas de investigación más activas, con aplicaciones en comprensión de imágenes, detección de objetos, reconocimiento facial, vigilancia de vídeo o procesamiento de imagen médica. La segmentación de imágenes es un problema difícil en general, pero especialmente en entornos científicos y biomédicos, donde las técnicas de adquisición imagen proporcionan imágenes ruidosas. Además, en muchos de estos casos se necesita una precisión casi perfecta. En esta tesis, revisamos y comparamos primero algunas de las técnicas ampliamente usadas para la segmentación de imágenes médicas. Estas técnicas usan clasificadores a nivel de pixel e introducen regularización sobre pares de píxeles que es normalmente insuficiente. Estudiamos las dificultades que presentan para capturar la información de alto nivel sobre los objetos a segmentar. Esta deficiencia da lugar a detecciones erróneas, bordes irregulares, configuraciones con topología errónea y formas inválidas. Para solucionar estos problemas, proponemos un nuevo método de regularización de alto nivel que aprende información topológica y de forma a partir de los datos de entrenamiento de una forma no paramétrica usando potenciales de orden superior. Los potenciales de orden superior se están popularizando en visión por computador, pero la representación exacta de un potencial de orden superior definido sobre muchas variables es computacionalmente inviable. Usamos una representación compacta de los potenciales basada en un conjunto finito de patrones aprendidos de los datos de entrenamiento que, a su vez, depende de las observaciones. Gracias a esta representación, los potenciales de orden superior pueden ser convertidos a potenciales de orden 2 con algunas variables auxiliares añadidas. Experimentos con imágenes reales y sintéticas confirman que nuestro modelo soluciona los errores de aproximaciones más débiles. Incluso con una regularización de alto nivel, una precisión exacta es inalcanzable, y se requeire de edición manual de los resultados de la segmentación automática. La edición manual es tediosa y pesada, y cualquier herramienta de ayuda es muy apreciada. Estas herramientas necesitan ser precisas, pero también lo suficientemente rápidas para ser usadas de forma interactiva. Los contornos activos son una buena solución: son buenos para detecciones precisas de fronteras y, en lugar de buscar una solución global, proporcionan un ajuste fino a resultados que ya existían previamente. Sin embargo, requieren una representación implícita que les permita trabajar con cambios topológicos del contorno, y esto da lugar a ecuaciones en derivadas parciales (EDP) que son costosas de resolver computacionalmente y pueden presentar problemas de estabilidad numérica. Presentamos una aproximación morfológica a la evolución de contornos basada en un nuevo operador morfológico de curvatura que es válido para superficies de cualquier dimensión. Aproximamos la solución numérica de la EDP de la evolución de contorno mediante la aplicación sucesiva de un conjunto de operadores morfológicos aplicados sobre una función de conjuntos de nivel. Estos operadores son muy rápidos, no sufren de problemas de estabilidad numérica y no degradan la función de los conjuntos de nivel, de modo que no hay necesidad de reinicializarlo. Además, su implementación es mucho más sencilla que la de las EDP, ya que no requieren usar sofisticados algoritmos numéricos. Desde un punto de vista teórico, profundizamos en las conexiones entre operadores morfológicos y diferenciales, e introducimos nuevos resultados en este área. Validamos nuestra aproximación proporcionando una implementación morfológica de los contornos geodésicos activos, los contornos activos sin bordes, y los turbopíxeles. En los experimentos realizados, las implementaciones morfológicas convergen a soluciones equivalentes a aquéllas logradas mediante soluciones numéricas tradicionales, pero con ganancias significativas en simplicidad, velocidad y estabilidad. ABSTRACT Image segmentation is an important field in computer vision and one of its most active research areas, with applications in image understanding, object detection, face recognition, video surveillance or medical image processing. Image segmentation is a challenging problem in general, but especially in the biological and medical image fields, where the imaging techniques usually produce cluttered and noisy images and near-perfect accuracy is required in many cases. In this thesis we first review and compare some standard techniques widely used for medical image segmentation. These techniques use pixel-wise classifiers and introduce weak pairwise regularization which is insufficient in many cases. We study their difficulties to capture high-level structural information about the objects to segment. This deficiency leads to many erroneous detections, ragged boundaries, incorrect topological configurations and wrong shapes. To deal with these problems, we propose a new regularization method that learns shape and topological information from training data in a nonparametric way using high-order potentials. High-order potentials are becoming increasingly popular in computer vision. However, the exact representation of a general higher order potential defined over many variables is computationally infeasible. We use a compact representation of the potentials based on a finite set of patterns learned fromtraining data that, in turn, depends on the observations. Thanks to this representation, high-order potentials can be converted into pairwise potentials with some added auxiliary variables and minimized with tree-reweighted message passing (TRW) and belief propagation (BP) techniques. Both synthetic and real experiments confirm that our model fixes the errors of weaker approaches. Even with high-level regularization, perfect accuracy is still unattainable, and human editing of the segmentation results is necessary. The manual edition is tedious and cumbersome, and tools that assist the user are greatly appreciated. These tools need to be precise, but also fast enough to be used in real-time. Active contours are a good solution: they are good for precise boundary detection and, instead of finding a global solution, they provide a fine tuning to previously existing results. However, they require an implicit representation to deal with topological changes of the contour, and this leads to PDEs that are computationally costly to solve and may present numerical stability issues. We present a morphological approach to contour evolution based on a new curvature morphological operator valid for surfaces of any dimension. We approximate the numerical solution of the contour evolution PDE by the successive application of a set of morphological operators defined on a binary level-set. These operators are very fast, do not suffer numerical stability issues, and do not degrade the level set function, so there is no need to reinitialize it. Moreover, their implementation is much easier than their PDE counterpart, since they do not require the use of sophisticated numerical algorithms. From a theoretical point of view, we delve into the connections between differential andmorphological operators, and introduce novel results in this area. We validate the approach providing amorphological implementation of the geodesic active contours, the active contours without borders, and turbopixels. In the experiments conducted, the morphological implementations converge to solutions equivalent to those achieved by traditional numerical solutions, but with significant gains in simplicity, speed, and stability.

Similarity in recognition processes used for bodies and faces: Bodies show caricature effects

Because faces and bodies share some abstract perceptual features, we hypothesised that similar recognition processes might be used for both. We investigated whether similar caricature effects to those found in facial identity and expression recognition could be found in the recognition of individual bodies and socially meaningful body positions. Participants were trained to name four body positions (anger, fear, disgust, sadness) and four individuals (in a neutral position). We then tested their recognition of extremely caricatured, moderately caricatured, anticaricatured, and undistorted images of each stimulus. Consistent with caricature effects found in face recognition, moderately caricatured representations of individuals' bodies were recognised more accurately than undistorted and extremely caricatured representations. No significant difference was found between participants' recognition of extremely caricatured, moderately caricatured, or undistorted body position line-drawings. AU anti-caricatured representations were named significandy less accurately than the veridical stimuli. Similar mental representations may be used for both bodies and faces.

Developmental trajectories of part-based and configural object recognition in adolescence.

Three experiments assessed the development of children's part and configural (part-relational) processing in object recognition during adolescence. In total, 312 school children aged 7-16 years and 80 adults were tested in 3-alternative forced choice (3-AFC) tasks. They judged the correct appearance of upright and inverted presented familiar animals, artifacts, and newly learned multipart objects, which had been manipulated either in terms of individual parts or part relations. Manipulation of part relations was constrained to either metric (animals, artifacts, and multipart objects) or categorical (multipart objects only) changes. For animals and artifacts, even the youngest children were close to adult levels for the correct recognition of an individual part change. By contrast, it was not until 11-12 years of age that they achieved similar levels of performance with regard to altered metric part relations. For the newly learned multipart objects, performance was equivalent throughout the tested age range for upright presented stimuli in the case of categorical part-specific and part-relational changes. In the case of metric manipulations, the results confirmed the data pattern observed for animals and artifacts. Together, the results provide converging evidence, with studies of face recognition, for a surprisingly late consolidation of configural-metric relative to part-based object recognition.

Coding Strategies and Implementations of Compressive Sensing

This dissertation studies the coding strategies of computational imaging to overcome the limitation of conventional sensing techniques. The information capacity of conventional sensing is limited by the physical properties of optics, such as aperture size, detector pixels, quantum efficiency, and sampling rate. These parameters determine the spatial, depth, spectral, temporal, and polarization sensitivity of each imager. To increase sensitivity in any dimension can significantly compromise the others.

This research implements various coding strategies subject to optical multidimensional imaging and acoustic sensing in order to extend their sensing abilities. The proposed coding strategies combine hardware modification and signal processing to exploiting bandwidth and sensitivity from conventional sensors. We discuss the hardware architecture, compression strategies, sensing process modeling, and reconstruction algorithm of each sensing system.

Optical multidimensional imaging measures three or more dimensional information of the optical signal. Traditional multidimensional imagers acquire extra dimensional information at the cost of degrading temporal or spatial resolution. Compressive multidimensional imaging multiplexes the transverse spatial, spectral, temporal, and polarization information on a two-dimensional (2D) detector. The corresponding spectral, temporal and polarization coding strategies adapt optics, electronic devices, and designed modulation techniques for multiplex measurement. This computational imaging technique provides multispectral, temporal super-resolution, and polarization imaging abilities with minimal loss in spatial resolution and noise level while maintaining or gaining higher temporal resolution. The experimental results prove that the appropriate coding strategies may improve hundreds times more sensing capacity.

Human auditory system has the astonishing ability in localizing, tracking, and filtering the selected sound sources or information from a noisy environment. Using engineering efforts to accomplish the same task usually requires multiple detectors, advanced computational algorithms, or artificial intelligence systems. Compressive acoustic sensing incorporates acoustic metamaterials in compressive sensing theory to emulate the abilities of sound localization and selective attention. This research investigates and optimizes the sensing capacity and the spatial sensitivity of the acoustic sensor. The well-modeled acoustic sensor allows localizing multiple speakers in both stationary and dynamic auditory scene; and distinguishing mixed conversations from independent sources with high audio recognition rate.

Development of Multi-modal and Super-resolved Retinal Imaging Systems

Advancements in retinal imaging technologies have drastically improved the quality of eye care in the past couple decades. Scanning laser ophthalmoscopy (SLO) and optical coherence tomography (OCT) are two examples of critical imaging modalities for the diagnosis of retinal pathologies. However current-generation SLO and OCT systems have limitations in diagnostic capability due to the following factors: the use of bulky tabletop systems, monochromatic imaging, and resolution degradation due to ocular aberrations and diffraction.

Bulky tabletop SLO and OCT systems are incapable of imaging patients that are supine, under anesthesia, or otherwise unable to maintain the required posture and fixation. Monochromatic SLO and OCT imaging prevents the identification of various color-specific diagnostic markers visible with color fundus photography like those of neovascular age-related macular degeneration. Resolution degradation due to ocular aberrations and diffraction has prevented the imaging of photoreceptors close to the fovea without the use of adaptive optics (AO), which require bulky and expensive components that limit the potential for widespread clinical use.

In this dissertation, techniques for extending the diagnostic capability of SLO and OCT systems are developed. These techniques include design strategies for miniaturizing and combining SLO and OCT to permit multi-modal, lightweight handheld probes to extend high quality retinal imaging to pediatric eye care. In addition, a method for extending true color retinal imaging to SLO to enable high-contrast, depth-resolved, high-fidelity color fundus imaging is demonstrated using a supercontinuum light source. Finally, the development and combination of SLO with a super-resolution confocal microscopy technique known as optical photon reassignment (OPRA) is demonstrated to enable high-resolution imaging of retinal photoreceptors without the use of adaptive optics.

The ENCARA system for face detection and normalization

[EN]Facial image processing is becoming widespread in human-computer applications, despite its complexity. High-level processes such as face recognition or gender determination rely on low-level routines that must e ectively detect and normalize the faces that appear in the input image. In this paper, a face detection and normalization system is described. The approach taken is based on a cascade of fast, weak classi ers that together try to determine whether a frontal face is present in the image.

A qualitative impairment in face perception in Alzheimer’s disease : evidence from a reduced face inversion effect

Prevalent face recognition difficulties in Alzheimer’s disease (AD) have typically been attributed to the underlying episodic and semantic memory impairment. The aim of the current study was to determine if AD patients are also impaired at the perceptual level for faces, more specifically at extracting a visual representation of an individual face. To address this question, we investigated the matching of simultaneously presented individual faces and of other nonface familiar shapes (cars), at both upright and inverted orientation, in a group of mild AD patients and in a group of healthy older controls matched for age and education. AD patients showed a reduced inversion effect (i.e., larger performance for upright than inverted stimuli) for faces, but not for cars, both in terms of error rates and response times. While healthy participants showed a much larger decrease in performance for faces than for cars with inversion, the inversion effect did not differ significantly for faces and cars in AD. This abnormal inversion effect for faces was observed in a large subset of individual patients with AD. These results suggest that AD patients have deficits in higher-level visual processes, more specifically at perceiving individual faces, a function that relies on holistic representations specific to upright face stimuli. These deficits, combined with their memory impairment, may contribute to the difficulties in recognizing familiar people that are often reported in patients suffering from the disease and by their caregivers.

Miglioramento della qualità di algoritmi di face morphing generation

L'obiettivo principale di questo lavoro di tesi è quello di migliorare gli algoritmi di morphing generation in termini di qualità visiva e di potenzialità di attacco dei sistemi automatici di riconoscimento facciale.

Quantum Dots As Biophotonics Tools.

This chapter provides a short review of quantum dots (QDs) physics, applications, and perspectives. The main advantage of QDs over bulk semiconductors is the fact that the size became a control parameter to tailor the optical properties of new materials. Size changes the confinement energy which alters the optical properties of the material, such as absorption, refractive index, and emission bands. Therefore, by using QDs one can make several kinds of optical devices. One of these devices transforms electrons into photons to apply them as active optical components in illumination and displays. Other devices enable the transformation of photons into electrons to produce QDs solar cells or photodetectors. At the biomedical interface, the application of QDs, which is the most important aspect in this book, is based on fluorescence, which essentially transforms photons into photons of different wavelengths. This chapter introduces important parameters for QDs' biophotonic applications such as photostability, excitation and emission profiles, and quantum efficiency. We also present the perspectives for the use of QDs in fluorescence lifetime imaging (FLIM) and Förster resonance energy transfer (FRET), so useful in modern microscopy, and how to take advantage of the usually unwanted blinking effect to perform super-resolution microscopy.

Amygdala hyperactivation in untreated depressed individuals

The amygdala participates in the detection and control of affective states, and has been proposed to be a site of dysfunction in affective disorders. To assess amygdala processing in individuals with unipolar depression, we applied a functional MRI (fMRI) paradigm previously shown to be sensitive to amygdala function. Fourteen individuals with untreated DSM-IV major depression and 15 healthy subjects were studied using fMRI with a standardized emotion face recognition task. Voxel-level data sets were subjected to a multiple-regression analysis, and functionally defined regions of interest (ROI), including bilateral amygdala, were analyzed with MANOVA. Pearson correlation coefficients between amygdala activation and HAM-D score also were performed. While both depressed and healthy groups showed increased amygdala activity when viewing emotive faces compared to geometric shapes, patients with unipolar depression showed relatively more activity than healthy subjects, particularly on the left. Positive Pearson correlations between amygdala activation and HAM-D score were found for both left and right ROIs in the patient group. This study provides in vivo imaging evidence to support the hypothesis of abnormal amygdala functioning in depressed individuals. (C) 2009 Elsevier Ireland Ltd. All rights reserved.

Three-dimensional space-borne synthetic aperture radar (SAR) imaging with multiple pass processing

Three-dimensional (3D) synthetic aperture radar (SAR) imaging via multiple-pass processing is an extension of interferometric SAR imaging. It exploits more than two flight passes to achieve a desired resolution in elevation. In this paper, a novel approach is developed to reconstruct a 3D space-borne SAR image with multiple-pass processing. It involves image registration, phase correction and elevational imaging. An image model matching is developed for multiple image registration, an eigenvector method is proposed for the phase correction and the elevational imaging is conducted using a Fourier transform or a super-resolution method for enhancement of elevational resolution. 3D SAR images are obtained by processing simulated data and real data from the first European Remote Sensing satellite (ERS-1) with the proposed approaches.

Dados de reconhecimento facial em Business Intelligence

Vivemos cada vez mais numa era de crescentes avanços tecnológicos em diversas áreas. O que há uns anos atrás era considerado como praticamente impossível, em muitos dos casos, já se tornou realidade. Todos usamos tecnologias como, por exemplo, a Internet, Smartphones e GPSs de uma forma natural. Esta proliferação da tecnologia permitiu tanto ao cidadão comum como a organizações a sua utilização de uma forma cada vez mais criativa e simples de utilizar. Além disso, a cada dia que passa surgem novos negócios e startups, o que demonstra o dinamismo que este crescimento veio trazer para a indústria. A presente dissertação incide sobre duas áreas em forte crescimento: Reconhecimento Facial e Business Intelligence (BI), assim como a respetiva combinação das duas com o objetivo de ser criado um novo módulo para um produto já existente. Tratando-se de duas áreas distintas, é primeiramente feito um estudo sobre cada uma delas. A área de Business Intelligence é vocacionada para organizações e trata da recolha de informação sobre o negócio de determinada empresa, seguindo-se de uma posterior análise. A grande finalidade da área de Business Intelligence é servir como forma de apoio ao processo de tomada de decisão por parte dos analistas e gestores destas organizações. O Reconhecimento Facial, por sua vez, encontra-se mais presente na sociedade. Tendo surgido no passado através da ficção científica, cada vez mais empresas implementam esta tecnologia que tem evoluído ao longo dos anos, chegando mesmo a ser usada pelo consumidor final, como por exemplo em Smartphones. As suas aplicações são, portanto, bastante diversas, desde soluções de segurança até simples entretenimento. Para estas duas áreas será assim feito um estudo com base numa pesquisa de publicações de autores da respetiva área. Desde os cenários de utilização, até aspetos mais específicos de cada uma destas áreas, será assim transmitido este conhecimento para o leitor, o que permitirá uma maior compreensão por parte deste nos aspetos relativos ao desenvolvimento da solução. Com o estudo destas duas áreas efetuado, é então feita uma contextualização do problema em relação à área de atuação da empresa e quais as abordagens possíveis. É também descrito todo o processo de análise e conceção, assim como o próprio desenvolvimento numa vertente mais técnica da solução implementada. Por fim, são apresentados alguns exemplos de resultados obtidos já após a implementação da solução.

Robust facial expression analysis for affect-based interaction

Dissertação para obtenção do Grau de Mestre em Engenharia Informática

Enhancing the performance of spread spectrum techniques in different applications

Spread spectrum, Automotive Radar, Indoor Positioning Systems, Ultrasonic and Microwave Imaging, super resolution technique and wavelet transform

Eigen Faces: Detecció, Classificació/Reconeixement de cares

En aquest projecte es pretén utilitzar mètodes coneguts com ara Viola&Jones (detecció) i EigenFaces (reconeixement) per a detectar i reconèixer cares dintre d’imatges de vídeo. Per a aconseguir aquesta tasca cal partir d’un conjunt de dades d’entrenament per a cada un dels mètodes (base de dades formada per imatges i anotacions manuals). A partir d’aquí, l’aplicació, ha de ser capaç de detectar cares en noves imatges i reconèixer-les (identificar de quina cara es tracta)