Parallel Computer Vision Algorithms for Graphics Processing Units

La evolución de los teléfonos móviles inteligentes, dotados de cámaras digitales, está provocando una creciente demanda de aplicaciones cada vez más complejas que necesitan algoritmos de visión artificial en tiempo real; puesto que el tamaño de las señales de vídeo no hace sino aumentar y en cambio el rendimiento de los procesadores de un solo núcleo se ha estancado, los nuevos algoritmos que se diseñen para visión artificial han de ser paralelos para poder ejecutarse en múltiples procesadores y ser computacionalmente escalables. Una de las clases de procesadores más interesantes en la actualidad se encuentra en las tarjetas gráficas (GPU), que son dispositivos que ofrecen un alto grado de paralelismo, un excelente rendimiento numérico y una creciente versatilidad, lo que los hace interesantes para llevar a cabo computación científica. En esta tesis se exploran dos aplicaciones de visión artificial que revisten una gran complejidad computacional y no pueden ser ejecutadas en tiempo real empleando procesadores tradicionales. En cambio, como se demuestra en esta tesis, la paralelización de las distintas subtareas y su implementación sobre una GPU arrojan los resultados deseados de ejecución con tasas de refresco interactivas. Asimismo, se propone una técnica para la evaluación rápida de funciones de complejidad arbitraria especialmente indicada para su uso en una GPU. En primer lugar se estudia la aplicación de técnicas de síntesis de imágenes virtuales a partir de únicamente dos cámaras lejanas y no paralelas—en contraste con la configuración habitual en TV 3D de cámaras cercanas y paralelas—con información de color y profundidad. Empleando filtros de mediana modificados para la elaboración de un mapa de profundidad virtual y proyecciones inversas, se comprueba que estas técnicas son adecuadas para una libre elección del punto de vista. Además, se demuestra que la codificación de la información de profundidad con respecto a un sistema de referencia global es sumamente perjudicial y debería ser evitada. Por otro lado se propone un sistema de detección de objetos móviles basado en técnicas de estimación de densidad con funciones locales. Este tipo de técnicas es muy adecuada para el modelado de escenas complejas con fondos multimodales, pero ha recibido poco uso debido a su gran complejidad computacional. El sistema propuesto, implementado en tiempo real sobre una GPU, incluye propuestas para la estimación dinámica de los anchos de banda de las funciones locales, actualización selectiva del modelo de fondo, actualización de la posición de las muestras de referencia del modelo de primer plano empleando un filtro de partículas multirregión y selección automática de regiones de interés para reducir el coste computacional. Los resultados, evaluados sobre diversas bases de datos y comparados con otros algoritmos del estado del arte, demuestran la gran versatilidad y calidad de la propuesta. Finalmente se propone un método para la aproximación de funciones arbitrarias empleando funciones continuas lineales a tramos, especialmente indicada para su implementación en una GPU mediante el uso de las unidades de filtraje de texturas, normalmente no utilizadas para cómputo numérico. La propuesta incluye un riguroso análisis matemático del error cometido en la aproximación en función del número de muestras empleadas, así como un método para la obtención de una partición cuasióptima del dominio de la función para minimizar el error. ABSTRACT The evolution of smartphones, all equipped with digital cameras, is driving a growing demand for ever more complex applications that need to rely on real-time computer vision algorithms. However, video signals are only increasing in size, whereas the performance of single-core processors has somewhat stagnated in the past few years. Consequently, new computer vision algorithms will need to be parallel to run on multiple processors and be computationally scalable. One of the most promising classes of processors nowadays can be found in graphics processing units (GPU). These are devices offering a high parallelism degree, excellent numerical performance and increasing versatility, which makes them interesting to run scientific computations. In this thesis, we explore two computer vision applications with a high computational complexity that precludes them from running in real time on traditional uniprocessors. However, we show that by parallelizing subtasks and implementing them on a GPU, both applications attain their goals of running at interactive frame rates. In addition, we propose a technique for fast evaluation of arbitrarily complex functions, specially designed for GPU implementation. First, we explore the application of depth-image–based rendering techniques to the unusual configuration of two convergent, wide baseline cameras, in contrast to the usual configuration used in 3D TV, which are narrow baseline, parallel cameras. By using a backward mapping approach with a depth inpainting scheme based on median filters, we show that these techniques are adequate for free viewpoint video applications. In addition, we show that referring depth information to a global reference system is ill-advised and should be avoided. Then, we propose a background subtraction system based on kernel density estimation techniques. These techniques are very adequate for modelling complex scenes featuring multimodal backgrounds, but have not been so popular due to their huge computational and memory complexity. The proposed system, implemented in real time on a GPU, features novel proposals for dynamic kernel bandwidth estimation for the background model, selective update of the background model, update of the position of reference samples of the foreground model using a multi-region particle filter, and automatic selection of regions of interest to reduce computational cost. The results, evaluated on several databases and compared to other state-of-the-art algorithms, demonstrate the high quality and versatility of our proposal. Finally, we propose a general method for the approximation of arbitrarily complex functions using continuous piecewise linear functions, specially formulated for GPU implementation by leveraging their texture filtering units, normally unused for numerical computation. Our proposal features a rigorous mathematical analysis of the approximation error in function of the number of samples, as well as a method to obtain a suboptimal partition of the domain of the function to minimize approximation error.

A class of algorithms for automatic evaluation of certain elementary functions in a binary computer.

"Supported in part...under Grant No. US NSF GJ812 and Grant No. US NSF GJ813."

On Jacobi and Jacobi-like algorithms for a parallel computer /

Includes bibliographies.

Process design and cost estimating algorithms for the computer assisted procedure for design and evaluation of wastewater treatment systems (CAPDET) /

"PB 82-190455."

The Computer System for the Estimation of the Realizability of the Crew Activity Algorithms

When designing specification on-board algorithm (the algorithm, realized on on-board digital computing machine, and algorithm to activity of the crew necessary to conduct the estimation their realizing. Presented computer system allows in interactive mode with user to value the temporary expenseses of the operator on processes decision making and their realizing, participations it in process of the spying.

Signal processing and information fusing algorithms for the synthesis of an alternative electromyogram/eye gaze tracking computer cursor control system

This research pursued the conceptualization and real-time verification of a system that allows a computer user to control the cursor of a computer interface without using his/her hands. The target user groups for this system are individuals who are unable to use their hands due to spinal dysfunction or other afflictions, and individuals who must use their hands for higher priority tasks while still requiring interaction with a computer. ^ The system receives two forms of input from the user: Electromyogram (EMG) signals from muscles in the face and point-of-gaze coordinates produced by an Eye Gaze Tracking (EGT) system. In order to produce reliable cursor control from the two forms of user input, the development of this EMG/EGT system addressed three key requirements: an algorithm was created to accurately translate EMG signals due to facial movements into cursor actions, a separate algorithm was created that recognized an eye gaze fixation and provided an estimate of the associated eye gaze position, and an information fusion protocol was devised to efficiently integrate the outputs of these algorithms. ^ Experiments were conducted to compare the performance of EMG/EGT cursor control to EGT-only control and mouse control. These experiments took the form of two different types of point-and-click trials. The data produced by these experiments were evaluated using statistical analysis, Fitts' Law analysis and target re-entry (TRE) analysis. ^ The experimental results revealed that though EMG/EGT control was slower than EGT-only and mouse control, it provided effective hands-free control of the cursor without a spatial accuracy limitation, and it also facilitated a reliable click operation. This combination of qualities is not possessed by either EGT-only or mouse control, making EMG/EGT cursor control a unique and practical alternative for a user's cursor control needs. ^

GOOAL: A routine based on Genetic Algorithms to track harmonics in electric power systems

Voltage and current waveforms of a distribution or transmission power system are not pure sinusoids. There are distortions in these waveforms that can be represented as a combination of the fundamental frequency, harmonics and high frequency transients. This paper presents a novel approach to identifying harmonics in power system distorted waveforms. The proposed method is based on Genetic Algorithms, which is an optimization technique inspired by genetics and natural evolution. GOOAL, a specially designed intelligent algorithm for optimization problems, was successfully implemented and tested. Two kinds of representations concerning chromosomes are utilized: binary and real. The results show that the proposed method is more precise than the traditional Fourier Transform, especially considering the real representation of the chromosomes.

Hybrid Algorithms for Routing and Assignment Wavelengths in Optical Networks

This paper presents a strategy for the solution of the WDM optical networks planning. Specifically, the problem of Routing and Wavelength Allocation (RWA) in order to minimize the amount of wavelengths used. In this case, the problem is known as the Min-RWA. Two meta-heuristics (Tabu Search and Simulated Annealing) are applied to take solutions of good quality and high performance. The key point is the degradation of the maximum load on the virtual links in favor of minimization of number of wavelengths used; the objective is to find a good compromise between the metrics of virtual topology (load in Gb/s) and of the physical topology (quantity of wavelengths). The simulations suggest good results when compared to some existing in the literature.

A Taxonomy for Locality Algorithms on Peer-to-Peer Networks

The continuous growth of peer-to-peer networks has made them responsible for a considerable portion of the current Internet traffic. For this reason, improvements in P2P network resources usage are of central importance. One effective approach for addressing this issue is the deployment of locality algorithms, which allow the system to optimize the peers` selection policy for different network situations and, thus, maximize performance. To date, several locality algorithms have been proposed for use in P2P networks. However, they usually adopt heterogeneous criteria for measuring the proximity between peers, which hinders a coherent comparison between the different solutions. In this paper, we develop a thoroughly review of popular locality algorithms, based on three main characteristics: the adopted network architecture, distance metric, and resulting peer selection algorithm. As result of this study, we propose a novel and generic taxonomy for locality algorithms in peer-to-peer networks, aiming to enable a better and more coherent evaluation of any individual locality algorithm.

Approximate algorithms for credal networks with binary variables

This paper presents a family of algorithms for approximate inference in credal networks (that is, models based on directed acyclic graphs and set-valued probabilities) that contain only binary variables. Such networks can represent incomplete or vague beliefs, lack of data, and disagreements among experts; they can also encode models based on belief functions and possibilistic measures. All algorithms for approximate inference in this paper rely on exact inferences in credal networks based on polytrees with binary variables, as these inferences have polynomial complexity. We are inspired by approximate algorithms for Bayesian networks; thus the Loopy 2U algorithm resembles Loopy Belief Propagation, while the Iterated Partial Evaluation and Structured Variational 2U algorithms are, respectively, based on Localized Partial Evaluation and variational techniques. (C) 2007 Elsevier Inc. All rights reserved.

Genetic algorithms for design of liquid retaining structure

In this paper, genetic algorithm (GA) is applied to the optimum design of reinforced concrete liquid retaining structures, which comprise three discrete design variables, including slab thickness, reinforcement diameter and reinforcement spacing. GA, being a search technique based on the mechanics of natural genetics, couples a Darwinian survival-of-the-fittest principle with a random yet structured information exchange amongst a population of artificial chromosomes. As a first step, a penalty-based strategy is entailed to transform the constrained design problem into an unconstrained problem, which is appropriate for GA application. A numerical example is then used to demonstrate strength and capability of the GA in this domain problem. It is shown that, only after the exploration of a minute portion of the search space, near-optimal solutions are obtained at an extremely converging speed. The method can be extended to application of even more complex optimization problems in other domains.

Use of the q-Gaussian mutation in evolutionary algorithms

This paper proposes the use of the q-Gaussian mutation with self-adaptation of the shape of the mutation distribution in evolutionary algorithms. The shape of the q-Gaussian mutation distribution is controlled by a real parameter q. In the proposed method, the real parameter q of the q-Gaussian mutation is encoded in the chromosome of individuals and hence is allowed to evolve during the evolutionary process. In order to test the new mutation operator, evolution strategy and evolutionary programming algorithms with self-adapted q-Gaussian mutation generated from anisotropic and isotropic distributions are presented. The theoretical analysis of the q-Gaussian mutation is also provided. In the experimental study, the q-Gaussian mutation is compared to Gaussian and Cauchy mutations in the optimization of a set of test functions. Experimental results show the efficiency of the proposed method of self-adapting the mutation distribution in evolutionary algorithms.