|
|
Unary Classification of Image Data
Beneš, Jiří ; Petyovský, Petr (referee) ; Horák, Karel (advisor)
The work deals with an introduction to classification algorithms. It then divides classifiers into unary, binary and multi-class and describes the different types of classifiers. The work compares individual classifiers and their areas of use. For unary classifiers, practical examples and a list of used architectures are given in the work. The work contains a chapter focused on the comparison of the effects of hyperparameters on the quality of unary classification for individual architectures. Part of the submission is a practical example of implementation of the unary classifier.
|
|
|
Unary Classification of Image Data
Beneš, Jiří ; Petyovský, Petr (referee) ; Horák, Karel (advisor)
The work deals with an introduction to classification algorithms. It then divides classifiers into unary, binary and multi-class and describes the different types of classifiers. The work compares individual classifiers and their areas of use. For unary classifiers, practical examples and a list of used architectures are given in the work. The work contains a chapter focused on the comparison of the effects of hyper parameters on the quality of unary classification for individual architectures. Part of the submission is a practical example of reimplementation of the unary classifier.
|
|
|
Deep neural networks and their application for image data processing
Golovizin, Andrey ; Mrázová, Iveta (advisor) ; Holan, Tomáš (referee)
In the area of image recognition, the so-called deep neural networks belong to the most promising models these days. They often achieve considerably better results than traditional techniques even without the necessity of any excessive task-oriented preprocessing. This thesis is devoted to the study and analysis of three basic variants of deep neural networks-namely the neocognitron, convolutional neural networks, and deep belief networks. Based on extensive testing of the described models on the standard task of handwritten digit recognition, the convolutional neural networks seem to be most suitable for the recognition of general image data. Therefore, we have used them also to classify images from two very large data sets-CIFAR-10 and ImageNet. In order to optimize the architecture of the applied networks, we have proposed a new pruning algorithm based on the Principal Component Analysis. Powered by TCPDF (www.tcpdf.org)
|
|
|
Maps creation from image data of the Mars surface
Kopecký, Kamil ; Kolář, Jan (advisor) ; Čábelka, Miroslav (referee)
Spectral behavior of Mars altimetry Abstract The goals of this thesis are research of cartography of Mars and the problems connected with it. The second goal is the creation of the maps of Mars's surface and lastly the analysis of the surface elevation and its spectral behavior. Firstly, the thesis outlines the theoretical part which includes usage of reference bodies, coordinates systems, cartographic projections and prime meridian. Second, it features the methodical procedure of creating the maps and the analysis. Key words: map projection, image data, Mars surface, thematic maps
|
|
|
Land-Cover classification of mountain ecosystem using image data with different spatial and spectral resolution
Minárčik, Miroslav ; Kupková, Lucie (advisor) ; Kříž, Jan (referee)
Land-Cover classification of mountain ecosystem using image data with different spatial and spectral resolution Abstract The bachelor thesis is focused on land cover classification of the western part of the Krkonoše Mts. using multispectral images from WorldView-2 satellite with spatial resolution 2 m and from Landsat 8 satellite with a spatial resolution 30 m. The goal was to compare results of supervised classifications Maximum Likelihod and Support Vector Machine and unsupervised classification ISODATA for both images. The best result was achieved for WorldView-2 image using Maximum Likelihood classification (overall accuracy 73,1 %). The best result for Landsat image was achieved using Support Vector Machine classification (overall accuracy 70,78 %). Keywords: Krkonoše Mts., WorldView-2, Landsat 8, classification, multispectral image
|
|
|
Ultrasound image processing
Pokorný, Vojtěch ; Harabiš, Vratislav (referee) ; Mézl, Martin (advisor)
Ultrasound is an important imaging modality used in medicine. This bachelor’s thesis includes theory regarding physical characteristics of ultrasonic waves, a brief overview of the historical development of ultrasound as an imaging modality in medicine, biophysics of spreading and biological effects of ultrasound waves on tissues and types of display final images. Next part of the thesis describes diagnostic ultrasound device, including the most common used types of ultrasound probes. This part is followed by description of animal Vevo 2100 ultrasound system from which the acquired data are processed further. Processing includes the loading, display and for DICOM files saving too. At various file formats DICOM standard comparison is done. The last chapter deals with the filtration of characteristic noise for ultrasound images called speckle.
|
|
|
Water Object Detection in Image
Čeloud, David ; Španěl, Michal (referee) ; Šilhavá, Jana (advisor)
Bachelor's thesis describes history of remote sensing, image data structure, their processing and analyzing. It defines mulstispectral space and explains basics of quantitative analysis and differences between supervised and unsupervised classification. Implementation section describes designing and developing of program, which will be able to open and process image data and detect water objects in him.
|
|
|
DICOM protocol analysis and its utilisation in image signal processing
Valouch, Lukáš ; Přinosil, Jiří (referee) ; Šmirg, Ondřej (advisor)
This Bachelor thesis describes DICOM which is a summary of standards for handling, storing and transmitting information in medical imaging. This standard is developed by organizations ACR and NEMA, it is very large file and it contains 18 of documents which are an actual version of this norm. The last norm was got out in 2008. The most important part of this format of date and files is data image processing so called off-line communication. It follows description of implemented application in C, C++ languages by the help of Imebra library. This Imebra library is free in terms of license of GNU GPL. The resultant program is standard of DICOM from, which is described above. This program is concerned with imaging of 16 bits of dept on 8 bits of dept of display devices, for example common computer displays. This application is able to display several layers of DICOM’s pictures. Each of layers contains separated bites which are presented by importance of levels.
|
|
|
High resolution computed tomography - its role and position in radiodiagnostics
CHYLÍKOVÁ, Renáta
In the introduction to my bachelor´s thesis, I have described the historical background of HRCT, its origins, and evolution. The historical part includes information about the development of CT equipment and the principles of diagnostics using the instruments that preceded HRCT. In the following part, I have described the principle of how HRCT works and the conditions that are needed for HRCT examination. I focused on lung examination and partly also on the examination of temporal bone. The objective of my thesis is to determine the frequency of CT and HRCT application, for the purposes of which I compared two leasing hospitals in the South Bohemian region. I issued from the hypothesis of a higher frequency of HRCT use. In my bachelor´s work I used foreign literature and sources and also the method of comparing data from two hospitals. The preferred method was the quantitative one, used for numerical statistics in the theoretical part; for the practical part I used the mentioned comparison of the number of examinations performed at two hospitals in South Bohemia. When confronting the two hospitals, a significantly higher frequency of HRCT use was found in Nemocnice Tábor a.s. This fact is explained by the use of Aquilion 16, which is a lower ? quality appliance, and Aquilion 64 in České Budějovice, as well as by the monitoring of patients with fibrosing processes (so ? called cordarone lung, etc.), and patients with unclear small lesions appearing on summing images when only HRCT examination was performed. HRCT is also performed as a part of routine examination of the lungs and mediastinum if the indication requires it, while check ? up examinations run only under the HRCT algorithm. In conclusion it can be stated that the difference between the examinations in hospital České Budějovice a.s. and Nemocnice Tábor a.s. is based on the difference between their equipment and also the slightly different approach to small lesions in the pulmonary parenchyma.The hypothesis that the HRCT examination is used more frequently than the common CT examination of the lungs and mediastinum was not confirmed, which results from the fact that the number od patients needing the examination of just small or larger changes in the pulmonary parenchyma is substantially lower than the number of patients with indicated classical examination of the lungs and mediastinum.
|
guest ::
