The aim of this thesis is to examine the possibility of a court detection program for sports videos that can identify the court even when some important elements are not visible. The study will also analyze what external factors may impact the program's accuracy in detecting all relevant elements.  These questions are answered through a combination of computer vision techniques and algorithms. The study utilizes Design Science Research (DSR) as its research methodology to develop an artifact. A dataset of padel sports videos are evaluated to measure the artifacts accuracy. The artifact utilizes multiple computer vision techniques from the OpenCV library to detect relevant lines and edges and project them onto the frame using a predetermined court model as reference.  The findings indicated that the developed artifact demonstrated a relatively consistent level of accuracy in court detection across multiple courts, whenever a detection was made. However, the frequency of successful detections exhibited some inconsistency. The research also found that external factors did not significantly influence the accuracy of court detection, yet they posed challenges to the program's overall consistency.

Submitted by Cristiane Chim (cristiane.chim@ucpel.edu.br) on 2016-10-17T11:10:28Z No. of bitstreams: 1 JULIO FRANCISCO ROCHA DE OLIVEIRA.pdf: 2087173 bytes, checksum: c38d1eda0641d13eef2ccdb59655b27a (MD5) / Made available in DSpace on 2016-10-17T11:10:28Z (GMT). No. of bitstreams: 1 JULIO FRANCISCO ROCHA DE OLIVEIRA.pdf: 2087173 bytes, checksum: c38d1eda0641d13eef2ccdb59655b27a (MD5) Previous issue date: 2016-08-01 / This work proposes the exploration of approximate adders circuits for the implementation of power-efficient for Image Processing. The Gaussian filter is a convolution operator which is used to blur images and to remove noise. On the other hand, the Gradient of an image measures how it is changing. Both blocks can be designed in hardware using only shifts and additions/subtractions. In this work we exploit a set of approximate adders in order to implement energy-efficient filters. The tree of adders of Gaussian and Gradient filters are implemented using one Copy of bits adder, as well as an Error-Tolerant Adders - ETA. The approximate architectures are compared to the best precise implementation of the filters. As the Gaussian and Gradient blocks are part of the Canny edge detector algorithm, we have implemented the tree of adders of the filters aiming this application. In particular, an algorithm was proposed in the scope of this work in order to achieve the best adder trees for the Gaussian and Gradient filters. The main results show that for an efficient power realization of this algorithm, the best strategy consists in the implementation of the Gaussian filter with ETA I adder, and the Gradient filter with the Copy of bits adder. The approximate Gaussian and Gradient filters were applied to the fully hardware of Canny edge detector. The main results showed that the approximate Canny edge detector architectures present the best performance and precision metrics results, for most of the cases, when using both the Copy of bits and ETA I adders. For these tests a set of true images were used. The synthesis results showed that the use of the Gaussian and Gradient filters including the Copy of bits and ETA I adders has been efficient to the hardwired Canny edge detector that presented both area and energy consumption reductions. / Este trabalho propõe a exploração de circuitos somadores aproximados para a implementação de filtros eficientes em consumo de potência para Processamento de Imagem. O filtro Gaussiano é um operador de convolução que é usado para borrar as imagens e remover ruídos. Por outro lado, o Gradiente de uma imagem quantifica o quanto uma imagem está mudando. Ambos os blocos podem ser implementados em hardware usando apenas operações de deslocamento e somas/subtrações. Nesse trabalho, um conjunto de somadores aproximados é explorado para a implementação de filtros eficientes em termos de energia. As árvores de somadores dos filtros Gaussiano e Gradiente são implementadas usando um somador aproximado baseado na cópia de bits para a saída, bem como somadores tolerantes a erros (ETA - Error-Tolerant Adders). As arquiteturas aproximadas são comparadas com as implementações dos filtros com somadores precisos. Como os blocos Gaussiano e Gradiente são partes integrantes do algoritmo de detecção de bordas de Canny, logo as árvores de somadores dos filtros Gaussiano e Gradiente foram implementadas visando a esta aplicação. Em particular, um algoritmo foi proposto no âmbito deste trabalho para encontrar a melhor composição da árvore de somadores nos filtros Gaussiano e Gradiente. Os principais resultados mostram que, para a realização eficiente em potência desse algoritmo, as melhores estratégias consistem na implementação do filtro Gaussiano com o somador ETA I e a implementação do filtro Gradiente com o somador baseado em cópia de bits. Os filtros Gaussiano e Gradiente aproximados foram aplicados ao circuito completo de detecção de bordas de Canny. Os resultados mostraram que as arquiteturas de detecção de bordas de Canny aproximadas, com somadores baseado na cópia de bits e ETAI, na maioria dos casos possuem melhores resultados em relação às métricas de desempenho e precisão, com relação à arquitetura precisa. Os testes foram realizados usando um conjunto de imagens reais. Os resultados da síntese em ASIC mostraram que, as aproximações dos filtros Gaussiano e Gradiente com os somadores baseado em cópia de bits e ETA I trazem economia em área e energia ao circuito de detecção de bordas de Canny.

Image segmentation is a process to identify regions of interest from digital images. Image segmentation plays an important role in medical image processing which enables a variety of clinical applications. It is also a tool to facilitate the detection of abnormalities such as cancerous lesions in the brain. Although numerous efforts in recent years have advanced this technique, no single approach solves the problem of segmentation for the large variety of image modalities existing today. Consequently, brain MRI segmentation remains a challenging task. The purpose of this thesis is to demonstrate brain MRI segmentation for delineation of tumors, ventricles and other anatomical structures using Insight Segmentation and Registration Toolkit (ITK) routines as the foundation. ITK is an open-source software system to support the Visible Human Project. Visible Human Project is the creation of complete, anatomically detailed, three-dimensional representations of the normal male and female human bodies. Currently under active development, ITK employs leading-edge segmentation and registration algorithms in two, three, and more dimensions. A goal of this thesis is to implement those algorithms to facilitate brain segmentation for a brain cancer research scientist.

Localization of Drosophila embryos in images is a fundamental step in an automatic computational system for the exploration of gene-gene interaction on Drosophila. Contour extraction of embryonic images is challenging due to many variations in embryonic images. In the thesis work, we develop a localization framework based on the analysis of connected components of edge pixels in a scale space. We propose criteria to select optimal scales for embryonic localization. Furthermore, we propose a scale mapping strategy to compress the range of a scale space in order to improve the efficiency of the localization framework. The effectiveness of the proposed framework and the scale mapping strategy are validated in our experiments.

Perception, hence the usage of visual sensors is indispensable in mobile and autonomous robotics. Visual sensors such as cameras, rigidly mounted on a robot frame are the most common usage scenario. In this case, the motion of the camera due to the motion of the moving platform as well as the resulting shocks or vibrations causes a number of distortions on video frame sequences. Two most important ones are the frame-to-frame changes of the line-of-sight (LOS) and the presence of motion blur in individual frames. The latter of these two, namely motion blur plays a particularly dominant role in determining the performance of many vision algorithms used in mobile robotics. It is caused by the relative motion between the vision sensor and the scene during the exposure time of the frame. Motion blur is clearly an undesirable phenomenon in computer vision not only because it degrades the quality of images but also causes other feature extraction procedures to degrade or fail. Although there are many studies on feature based tracking, navigation, object recognition algorithms in the computer vision and robotics literature, there is no comprehensive work on the effects of motion blur on different image features and their extraction. In this thesis, a survey of existing models of motion blur and approaches to motion deblurring is presented. We review recent literature on motion blur and deblurring and we focus our attention on motion blur induced degradation of a number of popular feature detectors. We investigate and characterize this degradation using video sequences captured by the vision system of a mobile legged robot platform. Harris Corner detector, Canny Edge detector and Scale Invariant Feature Transform (SIFT) are chosen as the popular feature detectors that are most commonly used for mobile robotics applications. The performance degradation of these feature detectors due to motion blur are categorized to analyze the effect of legged locomotion on feature performance for perception. These analysis results are obtained as a first step towards the stabilization and restoration of video sequences captured by our experimental legged robotic platform and towards the development of motion blur robust vision system.

Steered Response Power based algorithms are widely used for finding sound source location using microphone array systems. SRCP-PHAT is one such algorithm that has a robust performance under noisy and reverberant conditions. The algorithm creates a likelihood function over the field of view. This thesis employs image processing methods on SRCP-PHAT images, to exploit the difference in power levels and pixel patterns to discriminate between sound source and background pixels. Hough Transform based ellipse detection is used to identify the sound source locations by finding the centers of elliptical edge pixel regions typical of source patterns. Monte Carlo simulations of an eight microphone perimeter array with single and multiple sound sources are used to simulate the test environment and area under receiver operating characteristic (ROCA) curve is used to analyze the algorithm performance. Performance was compared to a simpler algorithm involving Canny edge detection and image averaging and an algorithms based simply on the magnitude of local maxima in the SRCP image. Analysis shows that Canny edge detection based method performed better in the presence of coherent noise sources.

The thesis deals with design of a program for entering various types of geometric objects in an image for the purpose of their further processing. The program should also contain algorithms to ease object entering (e.g. refining manually entered object position). In the first part there is a brief description of the computer vision and its basic methods used in the work as well as introduction of the OpenCV image processing library. The following part describes three types of geometric primitives that are implemented for now. Because the output of the program is in universal XML format, there is short chapter about the XML. After that, there are summarized some methods for searching of parametric description of geometric primitives in an image. The final chapter describes the proposed system and evaluates possibility and suitability of its usage for various types of images.

This thesis deals with real-time image processing including preprocessing, segmentation and classification of objects. On the basis of classification is determined rotation and position of objects. The aim of this project is to develop a modular application which will be able to monitor mobile units and determine their rotation and position in real time.

This thesis presents the Deriche edge detector as an interesting alternative to the commonly used edge detectors. The Deriche edge detector's design is presented to the reader as well as its strengths and weaknesses. Performance issues of the Deriche edge detector are described in comparison with the Canny edge detector together with recommendations for using the Deriche detector. Finally, edge detection quality of the Deriche edge detector is compared to the Canny edge detector using robust subjective evaluation method.

This thesis deals with curve detection in images. First, current methods used in this area of image processing are summarized and described. Main topic of this thesis is a comparison of methods of parametric curve detection, such as Hough transformation and RANSAC-based methods. These methods are compared according to several criteria which are the most important for precise edge detection.