
Organ segmentation
Kolomazník, Jan
Ì ØÐ ÇÖ Ò Ë Ñ ÒØ Ø ÓÒ ÙØ ÓÖ Â Ò ÃÓÐÓÑ ÞÒ Ô ÖØÑ ÒØ Ô ÖØÑ ÒØ Ó ËÓ ØÛ Ö Ò ÓÑÔÙØ Ö Ë Ò Ù Ø ÓÒ ËÙÔ ÖÚ ×ÓÖ ÊÆ Öž ÂÓ× È Ð Ò ËÙÔ ÖÚ ×ÓÖ³× ¹Ñ Ð Ö ×× ÂÓ× žÈ Ð ÒÑ ž ÙÒ ž Þ ÇÙÖ Ó Ð Û × ØÓ Ù× Ö ×ÙÐØ× ÔÖ × ÒØ Ò Å Öž Î Ð Ú ÃÖ ³× ÔÐÓÑ Ø × ×¸ Û Ó ÜÔ Ö Ñ Ø Û Ø ÓÖ Ò × Ñ ÒØ Ø ÓÒ Ð ÓÖ Ø Ñ×¸ Û Ò ÔÔÐ ØÓ ÚÓÐÙÑ Ñ ¹ ×ÙÖ Ñ ÒØ Ò Ì Ø ×Ô ÐÐÝ Ò Ý× Ò Ð Òž Ï ÔÔÐ Ø × Ö ×ÙÐØ× ÙÖ Ò Ú ÐÓÔ Ñ ÒØ Ó ÔÖ Ø ÐÐÝ ÔÐÓÝ Ð ÔÔÐ Ø ÓÒ¸ Ù Ð ÓÒ Ø Å Ú Ö Ñ ÛÓÖ ¸ Û ÔÖÓÚ × ÒØÙ Ø Ú ÒØ Ö ÓÖ ÒØ Ö Ø ÓÒ Û Ø × Ñ ÒØ Ø ÓÒ Ð ÓÖ Ø Ñ×ž Ï ÜØ Ò ÓÖ Ò Ð ¿ × Ñ ÒØ Ø ÓÒ Ð ÓÖ Ø Ñ¸ × ÓÒ ¾ ¹×ÔÐ Ò × Ñ ÒØ Ø ÓÒ Ñ Ø Ó ¸ Ý ×Ø Ø ×Ø Ð ÓÖ Ò ÑÓ Ð¸ Û × ÓÙÐ Ò Ö × ÖÓ Ù×ØÒ ×× Ó Ø Ñ ÒØ ÓÒ Ñ Ø Ó Ò Ö × ÒÙÑ Ö Ó Ø× Ô Ö Ñ Ø Ö×ž ÅÓ Ð Û × × Ò Ò ÓÖ Ö ØÓ Ö Ù× Ð Ò ÓØ Ö × Ñ ÒØ Ø ÓÒ Ð ÓÖ Ø Ñž Ã ÝÛÓÖ × ÚÓÐÙÑ Ø ¸ × Ñ ÒØ Ø ÓÒ¸ Ø Ú ÓÒØÓÙÖ×¸ Ô Ö Ñ ØÖ ×Ò ×¸ ÚÓÐÙÑ Ñ ¹ ×ÙÖ Ñ ÒØ¸ ÓÔØ Ñ Þ Ø ÓÒ¸ Ñ Ð Ñ Ò ¸ ¹×ÔÐ Ò × ¾


Interactive Processing of Volumetric Data
Kolomazník, Jan ; Pelikán, Josef (advisor) ; Czanner, Silvester (referee) ; Dokládal, Petr (referee)
Title: Interactive Processing of Volumetric Data Author: Jan Kolomazník Department: Department of Software and Computer Science Education Supervisor: RNDr. Josef Pelikán, Department of Software and Computer Science Education Abstract: Interactive visualization and segmentation of volumetric data are quite lim ited due to the increased complexity of the task and size of the input data in comparison to twodimensional processing. A special interactive segmentation workflow is presented, based on minimal graphcut search. The overall execution time was lowered by implementing all the computational steps on GPU, which required a design of massively parallel algorithms (using thousands of threads). To lower the computational burden even further the graph is constructed over the image subregions com puted by parallel watershed transformation. As a suitable formalism for a range of massively parallel algorithms was chosen cellular automata. A set of cellular automata extensions was defined, which allows efficient mapping and computation on GPU. Several variants of parallel watershed transformation are then defined in the form of cellular automaton. A novel form of 2D transfer function was presented, to improve direct volume visualization of the input data, suited for discriminating image features by their shape and...


Interactive Processing of Volumetric Data
Kolomazník, Jan ; Pelikán, Josef (advisor) ; Ďurikovič, Roman (referee) ; Dokládal, Petr (referee)
Title: Interactive Processing of Volumetric Data Author: Jan Kolomazník Department: Department of Software and Computer Science Education Supervisor: RNDr. Josef Pelikán, Department of Software and Computer Science Education Abstract: Interactive visualization and segmentation of volumetric data are quite lim ited due to the increased complexity of the task and size of the input data in comparison to twodimensional processing. A special interactive segmentation workflow is presented, based on minimal graphcut search. The overall execution time was lowered by implementing all the computational steps on GPU, which required a design of massively parallel algorithms (using thousands of threads). To lower the computational burden even further the graph is constructed over the image subregions com puted by parallel watershed transformation. As a suitable formalism for a range of massively parallel algorithms was chosen cellular automata. A set of cellular automata extensions was defined, which allows efficient mapping and computation on GPU. Several variants of parallel watershed transformation are then defined in the form of cellular automaton. A novel form of 2D transfer function was presented, to improve direct volume visualization of the input data, suited for discriminating image features by their shape and...

 

Methods of extending depth of field
Šťasta, Jakub ; Kolomazník, Jan (advisor) ; Dupej, Ján (referee)
Images taken with macro lenses, microscope lenses or lenses with large focal length suffer from shallow depth of field. It is possible to take several images of the scene and combine them artificially into one sharp image. In this work we review known approaches to exdended depth of field (EDF) and in detail discuss several chosen algorithms, both working in spatial domain and through wavelet transform. We implemented a framework for EDF algorithms in Java as an ImageJ plugin with all the chosen algorithms. Afterwards we compared their performance on both artificial and real datasets. Among the tests were also robustness to Gaussian and impulse noise. We also discuss the most common artifacts and disadvantages of the algorithms. 1


Animation of clouds
Šnábl, Jan ; Pelikán, Josef (advisor) ; Kolomazník, Jan (referee)
Mraky jsou atmosferický jev vznikající kondenzací vzdušné vlhkosti. Pro mraky je typický pomalý pohyb po obloze. Nafotografovaná sekvence snímků však při rychlém přehrání může působit trhaně. Cílem této práce je navrhnout a implementovat algoritmus na generování mezisnímků do timelapse sekvence snímků mraků za účelem plynulé animace. Tuto úlohu rozdělíme na dvě podúlohy. V první části se zaměříme na zjišťování pohybu v obrazu. K tomu zvolíme metodu založenou na korelaci. V druhé části se zaměříme na deformaci obrazu a výpočet výstupních mezisnímků na základě hodnot spočítaných v první fázi. V rámci této práce vytvoříme jednoduchý applet, na kterém lze demonstrovat výstupy v různých fázích výpočtu a vygenerovat sekvenci mezisnímků.

 

Optical Character Recognition at Camera Captured Images
Kindermann, Hubert ; Blažek, Jan (advisor) ; Kolomazník, Jan (referee)
We present solution of steps necessary for binarization and text lines detection contained in printed documents digitized by the camera. We introduce a normalization of nonuniform illumination method for text photographs. We propose input bitmap binarization algorithm based on twodimensional probability pixel model which also considers its surrounding. We continue with description of robust text lines orientation detector based on optimization of risk function using first order derivatives of image function. In the end we present text lines detection and segmentation algorithm. Final shape of segmented lines is optimized with usage of graph algorithm. Powered by TCPDF (www.tcpdf.org)


Multidimensional transfer functions for scalar volumetric data visualization
Mach, Pavel ; Kolomazník, Jan (advisor) ; Dupej, Ján (referee)
Direct volume rendering is an algorithm used for displaying threedimensional scalar data, like image from Computed Tomography. This algorithm makes use of a concept of Transfer functions for assigning optical properties to the data values. We studied two dimensional transfer functions, that besides primary values have additional dataset as an input. In particular, we studied computation of this secondary dataset with respect to the primary image function shape. This was done by analysing eigenvalues of the Hessian matrix in each image point. We proposed one formula and implemented several others for computing the probability that image point belongs to the blood vessel. Powered by TCPDF (www.tcpdf.org)


Haptic rendering for 6/3DOF haptic devices
Kadleček, Petr ; Kmoch, Petr (advisor) ; Kolomazník, Jan (referee)
Application of haptic devices expanded to fields like virtual manufacturing, virtual assembly or medical simulations. Advances in development of haptic devices have resulted in a wide distribution of assymetric 6/3DOF haptic devices. However, current haptic rendering algorithms work correctly only for symmetric devices. This thesis analyzes 3DOF and 6DOF haptic rendering algorithms and proposes an algorithm for 6/3DOF haptic rendering involving pseudohaptics. The 6/3DOF haptic rendering algorithm is implemented based on the previous analysis and tested in a user study.
