Local gist vision of man-made objects

Attention is usually modelled by sequential fixation of peaks in saliency maps. Those maps code local conspicuity: complexity, colour and texture. Such features have no relation to entire objects, unless also disparity and optical flow are considered, which often segregate entire objects from their background. Recently we developed a model of local gist vision: which types of objects are about where in a scene. This model addresses man-made objects which are dominated by a small shape repertoire: squares, rectangles, trapeziums, triangles, circles and ellipses. Only exploiting local colour contrast, the model can detect these shapes by a small hierarchy of cell layers devoted to low- and mid-level geometry. The model has been tested successfully on video sequences containing traffic signs and other scenes, and partial occlusions were not problematic.

Fast Geodesic Active Fields for Image Registration Based on Splitting and Augmented Lagrangian Approaches.

In this paper, we present an efficient numerical scheme for the recently introduced geodesic active fields (GAF) framework for geometric image registration. This framework considers the registration task as a weighted minimal surface problem. Hence, the data-term and the regularization-term are combined through multiplication in a single, parametrization invariant and geometric cost functional. The multiplicative coupling provides an intrinsic, spatially varying and data-dependent tuning of the regularization strength, and the parametrization invariance allows working with images of nonflat geometry, generally defined on any smoothly parametrizable manifold. The resulting energy-minimizing flow, however, has poor numerical properties. Here, we provide an efficient numerical scheme that uses a splitting approach; data and regularity terms are optimized over two distinct deformation fields that are constrained to be equal via an augmented Lagrangian approach. Our approach is more flexible than standard Gaussian regularization, since one can interpolate freely between isotropic Gaussian and anisotropic TV-like smoothing. In this paper, we compare the geodesic active fields method with the popular Demons method and three more recent state-of-the-art algorithms: NL-optical flow, MRF image registration, and landmark-enhanced large displacement optical flow. Thus, we can show the advantages of the proposed FastGAF method. It compares favorably against Demons, both in terms of registration speed and quality. Over the range of example applications, it also consistently produces results not far from more dedicated state-of-the-art methods, illustrating the flexibility of the proposed framework.

An Encoded Infrared Sheet Of Light Navigational Beacon System For Precise Localization Of Indoor Mobile Robot Vehicles

A new localization approach to increase the navigational capabilities and object manipulation of autonomous mobile robots, based on an encoded infrared sheet of light beacon system, which provides position errors smaller than 0.02m is presented in this paper. To achieve this minimal position error, a resolution enhancement technique has been developed by utilising an inbuilt odometric/optical flow sensor information. This system respects strong low cost constraints by using an innovative assembly for the digitally encoded infrared transmitter. For better guidance of mobile robot vehicles, an online traffic signalling capability is also incorporated. Other added features are its less computational complexity and online localization capability all these without any estimation uncertainty. The constructional details, experimental results and computational methodologies of the system are also described

Robot Motion Vision by Fixation

In many motion-vision scenarios, a camera (mounted on a moving vehicle) takes images of an environment to find the "motion'' and shape. We introduce a direct-method called fixation for solving this motion-vision problem in its general case. Fixation uses neither feature-correspondence nor optical-flow. Instead, spatio-temporal brightness gradients are used directly. In contrast to previous direct methods, fixation does not restrict the motion or the environment. Moreover, fixation method neither requires tracked images as its input nor uses mechanical tracking for obtaining fixated images. The experimental results on real images are presented and the implementation issues and techniques are discussed.

Learning-Based Approach to Real Time Tracking and Analysis of Faces

This paper describes a trainable system capable of tracking faces and facialsfeatures like eyes and nostrils and estimating basic mouth features such as sdegrees of openness and smile in real time. In developing this system, we have addressed the twin issues of image representation and algorithms for learning. We have used the invariance properties of image representations based on Haar wavelets to robustly capture various facial features. Similarly, unlike previous approaches this system is entirely trained using examples and does not rely on a priori (hand-crafted) models of facial features based on optical flow or facial musculature. The system works in several stages that begin with face detection, followed by localization of facial features and estimation of mouth parameters. Each of these stages is formulated as a problem in supervised learning from examples. We apply the new and robust technique of support vector machines (SVM) for classification in the stage of skin segmentation, face detection and eye detection. Estimation of mouth parameters is modeled as a regression from a sparse subset of coefficients (basis functions) of an overcomplete dictionary of Haar wavelets.

Visual Speech Synthesis by Morphing Visemes

We present MikeTalk, a text-to-audiovisual speech synthesizer which converts input text into an audiovisual speech stream. MikeTalk is built using visemes, which are a small set of images spanning a large range of mouth shapes. The visemes are acquired from a recorded visual corpus of a human subject which is specifically designed to elicit one instantiation of each viseme. Using optical flow methods, correspondence from every viseme to every other viseme is computed automatically. By morphing along this correspondence, a smooth transition between viseme images may be generated. A complete visual utterance is constructed by concatenating viseme transitions. Finally, phoneme and timing information extracted from a text-to-speech synthesizer is exploited to determine which viseme transitions to use, and the rate at which the morphing process should occur. In this manner, we are able to synchronize the visual speech stream with the audio speech stream, and hence give the impression of a photorealistic talking face.

A swarm intelligence method for feature tracking in AMV derivation

Feature tracking is a key step in the derivation of Atmospheric Motion Vectors (AMV). Most operational derivation processes use some template matching technique, such as Euclidean distance or cross-correlation, for the tracking step. As this step is very expensive computationally, often shortrange forecasts generated by Numerical Weather Prediction (NWP) systems are used to reduce the search area. Alternatives, such as optical flow methods, have been explored, with the aim of improving the number and quality of the vectors generated and the computational efficiency of the process. This paper will present the research carried out to apply Stochastic Diffusion Search, a generic search technique in the Swarm Intelligence family, to feature tracking in the context of AMV derivation. The method will be described, and we will present initial results, with Euclidean distance as reference.

Estimation of visual motion parameters used for bio-inspired navigation

Visual motion cues play an important role in animal and humans locomotion without the need to extract actual ego-motion information. This paper demonstrates a method for estimating the visual motion parameters, namely the Time-To-Contact (TTC), Focus of Expansion (FOE), and image angular velocities, from a sparse optical flow estimation registered from a downward looking camera. The presented method is capable of estimating the visual motion parameters in a complicated 6 degrees of freedom motion and in real time with suitable accuracy for mobile robots visual navigation.

Caminhamento fotogramétrico utilizando o fluxo óptico filtrado

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Estabilização postural do olhar: efeitos da distância do objeto

The aim of the present study was to analyze the effects of looking at targets located at different distances on body oscillation during tasks of distinct difficulties. In Experiment 1, ten participants in quiet stance fixated targets in three conditions: No object-far (fixation on far-target without near-target), Object-near (fixation on near target with fartarget), and Object-far (fixation on far-target with near-target). Mean oscillations of trunk in anterior-posterior axis were smallest in the Object-near condition; the No object-far and Object-far conditions were similar. In Experiment 2, seven participants in kiba-dachi, a karate stance, were submitted to three conditions: Blindfolded, No object-far, and Object-near. Mean oscillations of head and trunk in anterior-posterior axis were smaller in the Object-near as compared to blindfolded condition; trunk oscillated more during No object-far than Object-near condition. The results support the notion that a simple posture is not automatically regulated by the optical flow, but different amounts of visual instability may be tolerated according to the fixation distance, regardless the presence of non-fixated objects; the control of a more difficult posture may also accommodate the effects of fixation distance.

Compact and autonomous multiwavelength microanalyzer for in-line and in situ colorimetric determinations

Nowadays, the attainment of microsystems that integrate most of the stages involved in an analytical process has raised an enormous interest in several research fields. This approach provides experimental set-ups of increased robustness and reliability, which simplify their application to in-line and continuous biomedical and environmental monitoring. In this work, a novel, compact and autonomous microanalyzer aimed at multiwavelength colorimetric determinations is presented. It integrates the microfluidics (a three-dimensional mixer and a 25 mm length "Z-shape" optical flow-cell), a highly versatile multiwavelength optical detection system and the associated electronics for signal processing and drive, all in the same device. The flexibility provided by its design allows the microanalyzer to be operated either in single fixed mode to provide a dedicated photometer or in multiple wavelength mode to obtain discrete pseudospectra. To increase its reliability, automate its operation and allow it to work under unattended conditions, a multicommutation sub-system was developed and integrated with the experimental set-up. The device was initially evaluated in the absence of chemical reactions using four acidochromic dyes and later applied to determine some key environmental parameters such as phenol index, chromium(VI) and nitrite ions. Results were comparable with those obtained with commercial instrumentation and allowed to demonstrate the versatility of the proposed microanalyzer as an autonomous and portable device able to be applied to other analytical methodologies based on colorimetric determinations.

Estimación del movimiento de objetos en secuencias de imágenes de intensidad y profundidad

Máster Universitario en Sistemas Inteligentes y Aplicaciones Numéricas en Ingeniería (SIANI)

3D reconstruction from a vascular tree model

[EN] In this paper, we present a vascular tree model made with synthetic materials and which allows us to obtain images to make a 3D reconstruction.We have used PVC tubes of several diameters and lengths that will let us evaluate the accuracy of our 3D reconstruction. In order to calibrate the camera we have used a corner detector. Also we have used Optical Flow techniques to follow the points through the images going and going back. We describe two general techniques to extract a sequence of corresponding points from multiple views of an object. The resulting sequence of points will be used later to reconstruct a set of 3D points representing the object surfaces on the scene. We have made the 3D reconstruction choosing by chance a couple of images and we have calculated the projection error. After several repetitions, we have found the best 3D location for the point.

Model of vessels for the 3D reconstruction

[EN] In this paper, we present a vascular tree model made with synthetic materials and which allows us to obtain images to make a 3D reconstruction. In order to create this model, we have used PVC tubes of several diameters and lengths that will let us evaluate the accuracy of our 3D reconstruction. We have made the 3D reconstruction from a series of images that we have from our model and after we have calibrated the camera. In order to calibrate it we have used a corner detector. Also we have used Optical Flow techniques to follow the points through the images going and going back. Once we have the set of images where we have located a point, we have made the 3D reconstruction choosing by chance a couple of images and we have calculated the projection error. After several repetitions, we have found the best 3D location for the point.

Studio ed implementazione di algoritmi per la navigazione inerziale di velivoli autonomi ad ala rotante

I sistemi di navigazione inerziale, denominati INS, e quelli di navigazione inerziale assistita, ovvero che sfruttano anche sensori di tipo non inerziale come ad esempio il GPS, denominati in questo caso INS/GPS, hanno visto un forte incremento del loro utilizzo soprattutto negli ultimi anni. I filtri complementari sfruttano segnali in ingresso che presentano caratteristiche complementari in termine di banda. Con questo lavoro di tesi mi sono inserito nel contesto del progetto SHERPA (Smart collaboration between Humans and ground-aErial Robots for imProving rescuing activities in Alpine environments), un progetto europeo, coordinato dall'Università di Bologna, che prevede di mettere a punto una piattaforma robotica in grado di aiutare i soccorritori che operano in ambienti ostili, come quelli del soccorso alpino, le guardie forestali, la protezione civile. In particolare è prevista la possibilità di lanciare i droni direttamente da un elicottero di supporto, per cui potrebbe essere necessario effettuare l'avvio del sistema in volo. Ciò comporta che il sistema di navigazione dovrà essere in grado di convergere allo stato reale del sistema partendo da un grande errore iniziale, dal momento che la fase di inizializzazione funziona bene solo in condizioni di velivolo fermo. Si sono quindi ricercati, in special modo, schemi che garantissero la convergenza globale. Gli algoritmi implementati sono alla base della navigazione inerziale, assistita da GPS ed Optical Flow, della prima piattaforma aerea sviluppata per il progetto SHERPA, soprannominata DreamDroneOne, che include una grande varietà di hardware appositamente studiati per il progetto, come il laser scanner, la camera termica, ecc. Dopo una panoramica dell'architettura del sistema di Guida, Navigazione e Controllo (GNC) in cui mi sono inserito, si danno alcuni cenni sulle diverse terne di riferimento e trasformazioni, si descrivono i diversi sensori utilizzati per la navigazione, si introducono gli AHRS (Attitude Heading Rference System), per la determinazione del solo assetto sfruttando la IMU ed i magnetometri, si analizza l'AHRS basato su Extended Kalman Filter. Si analizzano, di seguito, un algoritmo non lineare per la stima dell'assetto molto recente, e il sistema INS/GPS basato su EKF, si presenta un filtro complementare molto recente per la stima di posizione ed assetto, si presenta un filtro complementare per la stima di posizione e velocità, si analizza inoltre l'uso di un predittore GPS. Infine viene presentata la piattaforma hardware utilizzata per l'implementazione e la validazione, si descrive il processo di prototipazione software nelle sue fasi e si mostrano i risultati sperimentali.