National Repository of Grey Literature 71 records found  1 - 10nextend  jump to record: Search took 0.00 seconds. 
Tree trunk detection and recognition in images
Šalomon, Filip ; Hoderová, Jana (referee) ; Procházková, Jana (advisor)
This thesis deals with detection of circular shapes in images -- forest scan cross-sections. Freeman chain code algorithm is used for image segmentation. Ransac (Random Sample Consensus) algorithm is used for diameter measurement. The designed algorithm is tested on data from Žofín Forest.
Audio signal denoising using deep learning
Pacal, Tomáš ; Záviška, Pavel (referee) ; Mokrý, Ondřej (advisor)
This thesis deals with noise removal in audio signal using deep learning. The basic types of neural networks and their use in audio signal processing are described. The possibilities of implementing neural networks are tested in Matlab and Python. Subsequently, a~convolutional neural network model is proposed, according to which four different convolutional network architectures are implemented and then trained and tested on different types of noise. Based on these tests, one architecture was selected and subjected to a comparative test on a speech recording and then on a music recording, together with a noise reduction method using wavelet transform. The results are evaluated using both objective sound quality metrics and an informal listening test. The neural network achieved better results according to all the metrics used as well as in the listening test.
Physical aspects of digital image and object detection using convolutional neural networsk
The thesis is focused on computer vision and the comparison of trained deep learning models. In the theoretical part, a detailed overview of the physical properties of the visible electromagnetic spectrum and the principles of the digital image, including it, is processed. This is followed by a review of image processing methods, where conventional methods and then convolutional neural networks are briefly characterized. The practical part is focused on creating a suitable dataset with annotations, which is further applied for training selected models. The variability and accuracy of the obtained results is analyzed from the point of view of the selected evaluation metrics, as well as based on the experimental determination of input parameters for training.
Vývoj softwaru pro identifikaci střetu ptactva se skleněnou překážkou
The thesis is focused on the issue of image processing, its output is software for detecting changes in the image with implemented basic algorithms of this area, which is created in the programming language MATLAB. The first part is divided into three chapters and oriented on the theoretical introduction of the topic. The first chapter deals with linear filters of the spatial domain and their convolution with the image. The second chapter briefly describes the operations in the frequency domain, the Fourier transform, and the basic frequency filters, and the third deals with image segmen-tation with a focus on edge detection. The second part of the thesis includes an intro-duction to the development of the program, including its block diagram, pseudocode of applied algorithms with its short description, and an approach to the creation of a graphical user environment.
Fourier transform of periodic structures
Zajíc, Tomáš ; Zahradník, Miloš (advisor) ; Krýsl, Svatopluk (referee)
Mathematical description of Fourier transform of the periodic structure. We introduce the concept of the Fourier series and we investigate the Dirichlet kernel. We also introduce the concept of distributions, the Fourier transform and convolution. Using this we discover the properties of the Dirac's delta, the Dirac comb and then we define the periodic structure. In conclusion, we mention the dual lattice. The thesis is designed to contain physical notes. Some of proofs are formal.
Comparison Of Discretization Methods
Kárský, Vilém
This paper deals with discretization of the continuous systems. There will be presented two common methods how to do this job and one uncommon. The uncommon method is to look at the system as filter. So the system could be implemented as a FIR filter. In the end of this paper these methods will be compared.
Atrial Fibrillation Classification Using Deep Convolution Networks
Novotna, Petra
We propose the usage of three deep convolutional neural networks architectures for classification of a single lead electrocardiogram (ECG) recordings and evaluate them on the atrial fibrillation (AFIB) classification, for which data set was provided by the Department of Biomedical Engineering, BUT. The compared networks are based on ResNet, VGG net and AlexNet. Single lead signals are transformed into the form of spectrogram. AFIB data was augmented for the purpose of similar size of both respected classes and for successful classification. The most successful architecture, based on AlexNet, was found to perform obtaining an accuracy of 92 % and F1 score of 56 % on the hidden testing set.
High dynamic range images visualization methods
Markovsky, Aleksander ; Kosová, Petra (referee) ; Druckmüller, Miloslav (advisor)
This thesis describes image processing of pictures with high dynamic range. The first part aims to describe ways to reduce visible noise, Fourier transform and convolution for image processing, linear filtering and unsharp mask method. Theoretical framework is followed by a C++ program, that processes HDR images, enhances details and creates a .jpg output.
Application of computational methods in classification of glass stones
Lébl, Matěj ; Hnětynková, Iveta (advisor)
Application of computational methods in classification of glass stones Bc. Matěj Lébl Abstrakt: The goal of this thesis is to employ mathematical image processing methods in automatic quality control of glass jewellery stones. The main math- ematical subject is a matrix of specific attributes representing digital image of the studied products. First, the thesis summarizes mathematical definition of digital image and some standard image processing methods. Then, a complete solution to the considered problem is presented. The solution consists of stone localization within the image followed by analysis of the localized area. Two lo- calization approaches are presented. The first is based on the matrix convolution and optimized through the Fourier transform. The second uses mathematical methods of thresholding and median filtering, and data projection into one di- mension. The localized area is analyzed based on statistical distribution of the stone brightness. All methods are implemented in the MATLAB environment. 1
Object tracking in high-speed camera images
Myška, Michal ; Druckmüller, Miloslav (referee) ; Štarha, Pavel (advisor)
This master thesis is dealing with object tracking in high-speed camera images, within what we are trying to find their trajectory and orientation. The mathematical theory associated with this problem as well as the methods used fo image processing are described here. The main outcome is an application with a user interface through which we can calculate the desired parameters of the individual objects.

National Repository of Grey Literature : 71 records found   1 - 10nextend  jump to record: