|
|
Detekce a reprezentace hran v mračnech bodů
Večeřová, Alena ; Druckmüllerová, Hana (oponent) ; Procházková, Jana (vedoucí práce)
Tato práce se zabývá detekcí a vhodnou reprezentací hran v mračnech bodů. Pro detekci hranových bodů je využita metoda clusterování normál. Při reprezentaci přímkových hran je využita metoda RANSAC. V rámci práce jsou hranové body jednoho modelu proloženy i kružnicí za využití metody nejmenších čtverců a zobrazení v projektivním prostoru. Všechny metody byly implementovány v programu MATLAB a byly otestovány na reálných datech získaných na skeneru ATOS CompactScan 2M.
|
|
|
Application of Adaptive Filters in Processing of Solar Corona Images
Druckmüllerová, Hana ; Horová, Ivana (oponent) ; Rušin, Vojtech (oponent) ; Martišek, Dalibor (vedoucí práce)
Solar corona photography counts among the most complicated tasks in astrophotography. It also plays a key role for research of the solar corona. This thesis brings an a complete overview of methods for imaging the solar corona. The thesis contains necessary methematical background, the sequence of steps for image processing, an overview of adaptive filters used for visualization of corona structures in digital images, and new methods are proposed, especially for images which contain more noise than it is typical for images of the white corona taken during total solar eclipses, e.g. images taken with narrow-band filters. The Fourier normalizing-radial-graded filter method that I proposed during my PhD study are based on approximation of pixel values and their variability with trigonometric polynomials using other properties of the image.
|
|
|
Phase-correlation based image registration
Druckmüllerová, Hana ; Martišek, Dalibor (oponent) ; Procházková, Jana (vedoucí práce)
This thesis deals with the use of phase correlation for estimation of mutual rotation, scale-change and translation between digital images. Phase correlation is based on the Fourier transform, therefore the thesis describes the Fourier transform of functions defined on R^2 and the discrete Fourier transform of functions defined on a finite number of points {0, 1, ... ,N-1}^2, where N is a natural number. The thesis describes modifications of the phase correlations for estimation of similarity transformation parameters between images with high dynamic range and faint structures, containing additive or impulse noise and images which are taken with different sensors and optical systems. The thesis also focuses on the modifications of the method used for registration of solar corona images taken during total solar eclipses, which counts among the hardest registration tasks.
|
|
| |
|
|
A method for the visualization of high phase gradients in a microscopic image
Druckmüllerová, Hana ; Martišek, Dalibor (oponent) ; Chmelík, Radim (vedoucí práce)
Holographic microscopy is an unconventional microscopy technique suitable especially for transparent samples. It enables to visualize the refractive index of observed objects. A unique transmitted-light digital holographic microscope (TDHM) has been constructed at Institute of Physical Engineering, Faculty of Mechanical Engineering, Brno University of Technology. Holograms captured by the microscope are processed by means of a technique based on the Fourier transform in order to reconstruct the intensity and phase of the light waves passing through the observed object. The phase describes the object refractive index and thickness. In places where the object refractive index or thickness changes, also the phase does. The task of this bachelor thesis was to find a method for visualizing places with high phase gradient. A gradient computation method which was created does not require phase unwrapping and is therefore suitable generally for any images. The method was implemented in a computer software called Gradient3D, which enables not only to compute the phase gradient in two and three dimensions, but also to create color images composed from combinations of intensity, phase and gradient. It also contains methods for handling places with low reconstructed intensity where the phase value is unreliable and usually causes false phase gradients. The program has been tested on several image sets from the TDHM capturing biological specimens.
|
|
|
Image Descriptors and their Usage for Object Detection
Bature, Jonathan ; Druckmüllerová, Hana (oponent) ; Procházková, Jana (vedoucí práce)
Face detection in images is a widely explored topic in computer vision. The algorithm that enabled face recognition and set new standards in this area was the Viola-Jones algorithm. This thesis work describes and explain the actual implementation of a face detector based on the Viola Jones algorithm using the Matlab CascadeObjectDetector. This thesis contributes to the exploration of techniques such as principal component analysis (PCA) for dimensional reduction of descriptors to establish an object detection system that achieves the best trade-off between performance and speed. In our approach, face detection is performed by Principal Component Analysis (PCA). The facial image is projected into the facial space that encodes the best variation of known facial images. The space of the face is defined by the eigenface. An eigenface is a set of facial eigenvectors that may not correspond to common facial features such as eyes, nose, and lips. The system works by projecting pre-extracted facial images into a series of facial spaces that represent large deviations between known facial images. Faces are classified as known or unknown faces after matching with an existing face image on the database.
|
|
|
Boundary effects in signal processing: From classical problems to new opportunities
Popoola, Seyi James ; Druckmüllerová, Hana (oponent) ; Cicone, Antonio (vedoucí práce)
This thesis focuses on boundary issues in the decomposition of signals. Classical methods boundary issues have been studied extensively, but a new generation of methods has been introduced in the last couple of decades. The implementation of these novel methods has the potential to produce in an efficient way more accurate, flexible, and interpretable results, which can help advance research in real-life applications in various fields. Boundary issues for these techniques have been studied only recently. The results published so far shows that these methods have limitations and assumptions that need to be carefully considered to avoid potential misuse. In this study, we pinpoint and tackle the major obstacles associated with the use these new methods, and the recommendations on how to utilize them most effectively is also provided. To further illustrate the potential consequences of improper usage, we undertake a comprehensive examination of their application to actual data and carry out numerical simulations. Lastly, we propose a set of best practices for optimizing the performance of these techniques in the context of signal decomposition. A crucial suggestion is to employ, before applying any decomposition method, a signal extension technique as a means of mitigating boundary effects.
|
|
|
Fourierovské metody omezení optických artefaktů ve snímcích z SOHO EIT
Satýnek, Daniel ; Kosová, Petra (oponent) ; Druckmüllerová, Hana (vedoucí práce)
Tato práce se zabývá aplikací dvourozměrné diskrétní Fourierovy transformace pro zpracování a analýzu obrazových dat z vesmírného teleskopu. Bude popsána metoda, která slouží k eliminaci periodických artefaktů způsobených úzkopásmovými filtry na snímcích z EIT sondy SOHO. Diskutován je návrh filtru ve frekvenční oblasti, který slouží k potlačení periodických struktur v obraze. Navržená metoda dopočítává amplitudy filtrovaných frekvencí na základě aproximace okolních amplitud. Bude popsán algoritmus metody a jak nastavit její parametry.
|
|
|
Image Descriptors and their Usage for Object Detection
Bature, Jonathan ; Druckmüllerová, Hana (oponent) ; Procházková, Jana (vedoucí práce)
Face detection in images is a widely explored topic in computer vision. The algorithm that enabled face recognition and set new standards in this area was the Viola-Jones algorithm. This thesis work describes and explain the actual implementation of a face detector based on the Viola Jones algorithm using the Matlab CascadeObjectDetector. This thesis contributes to the exploration of techniques such as principal component analysis (PCA) for dimensional reduction of descriptors to establish an object detection system that achieves the best trade-off between performance and speed. In our approach, face detection is performed by Principal Component Analysis (PCA). The facial image is projected into the facial space that encodes the best variation of known facial images. The space of the face is defined by the eigenface. An eigenface is a set of facial eigenvectors that may not correspond to common facial features such as eyes, nose, and lips. The system works by projecting pre-extracted facial images into a series of facial spaces that represent large deviations between known facial images. Faces are classified as known or unknown faces after matching with an existing face image on the database.
|
|
|
Application of Adaptive Filters in Processing of Solar Corona Images
Druckmüllerová, Hana ; Horová, Ivana (oponent) ; Rušin, Vojtech (oponent) ; Martišek, Dalibor (vedoucí práce)
Solar corona photography counts among the most complicated tasks in astrophotography. It also plays a key role for research of the solar corona. This thesis brings an a complete overview of methods for imaging the solar corona. The thesis contains necessary methematical background, the sequence of steps for image processing, an overview of adaptive filters used for visualization of corona structures in digital images, and new methods are proposed, especially for images which contain more noise than it is typical for images of the white corona taken during total solar eclipses, e.g. images taken with narrow-band filters. The Fourier normalizing-radial-graded filter method that I proposed during my PhD study are based on approximation of pixel values and their variability with trigonometric polynomials using other properties of the image.
|
host ::
RSS.