This thesis compares hand-designed features with features learned by feature learning methods in video classification. The features learned by Principal Component Analysis whitening, Independent subspace analysis and Sparse Autoencoders were tested in a standard Bag of Visual Word classification paradigm replacing hand-designed features (e.g. SIFT, HOG, HOF). The classification performance was measured on Human Motion DataBase and YouTube Action Data Set. Learned features showed better performance than the hand-desined features. The combination of hand-designed features and learned features by Multiple Kernel Learning method showed even better performance, including cases when hand-designed features and learned features achieved not so good performance separately.

This thesis is dedicated to techniques for motion detection in video sequences. There is summary of several possible approaches to solve this matter in theoretical and practical point of view. Principles of motion detection based on difference between frames are discussed in details. Great attention is made for image description by new method Local Binary Patterns which allows creating of incredibly fast and accurate motion detector. Vital part of this thesis is application for monitoring and testing motion algorithms on video sequences in different environments. Results of measuring enables to determine best approaches for concrete place and reader might find it useful in case of designing his own motion detectors.

<p>У овој дисертацији разматране су објективне мере процене квалитета<br />слике и видеа са потпуним и делимичним референцирањем на изворни<br />сигнал. За потребе евалуације квалитета развијене су поуздане,<br />рачунски ефикасне мере, засноване на очувању информација о<br />градијенту. Мере су тестиране на великом броју тест слика и видео<br />секвенци, различитих типова и степена деградације. Поред јавно<br />доступних база слика и видео секвенци, за потребе истраживања<br />формиране су и нове базе видео секвенци са преко 300 релевантних<br />тест узорака. Поређењем доступних субјективних и објективних скорова<br />квалитета показано је да је објективна евалуација квалитета веома<br />сложен проблем, али га је могуће решити и доћи до високих<br />перформанси коришћењем предложених мера процене квалитета слике<br />и видеа.</p> / <p>U ovoj disertaciji razmatrane su objektivne mere procene kvaliteta<br />slike i videa sa potpunim i delimičnim referenciranjem na izvorni<br />signal. Za potrebe evaluacije kvaliteta razvijene su pouzdane,<br />računski efikasne mere, zasnovane na očuvanju informacija o<br />gradijentu. Mere su testirane na velikom broju test slika i video<br />sekvenci, različitih tipova i stepena degradacije. Pored javno<br />dostupnih baza slika i video sekvenci, za potrebe istraživanja<br />formirane su i nove baze video sekvenci sa preko 300 relevantnih<br />test uzoraka. Poređenjem dostupnih subjektivnih i objektivnih skorova<br />kvaliteta pokazano je da je objektivna evaluacija kvaliteta veoma<br />složen problem, ali ga je moguće rešiti i doći do visokih<br />performansi korišćenjem predloženih mera procene kvaliteta slike<br />i videa.</p> / <p>This thesis presents an investigation into objective image and video quality<br />assessment with full and reduced reference on original (source) signal. For<br />quality evaluation purposes, reliable, computational efficient, gradient-based<br />measures are developed. Proposed measures are tested on different image<br />and video datasets, with various types of distorsions and degradation levels.<br />Along with publicly available image and video quality datasets, new video<br />quality datasets are maded, with more than 300 relevant test samples.<br />Through comparison between available subjective and objective quality<br />scores it has been shown that objective quality evaluation is highly complex<br />problem, but it is possible to resolve it and acchieve high performance using<br />proposed quality measures.</p>

This master's thesis Multimedia support of teaching mathematics aims at supporting teachers in introducing innovations in teaching and support in their own teaching materials. It is dedicated to supporting the pupil in self-education. It also serves as an overview of selected resources to support the use of technology and software in teaching. This master's thesis includes examples of the use of mathematical programs in mathematics classes, and a proposal to engage pupils in teaching, by implementing the "Teach Your Fellow" project, which aims to create learning videos for pupils.

Title: Image Deblurring in Demanding Conditions Author: Jan Kotera Department: Institute of Information Theory and Automation, Czech Academy of Sciences Supervisor: Doc. Ing. Filip Šroubek, Ph.D., DSc., Institute of Information Theory and Automation, Czech Academy of Sciences Abstract: Image deblurring is a computer vision task consisting of removing blur from image, the objective is to recover the sharp image corresponding to the blurred input. If the nature and shape of the blur is unknown and must be estimated from the input image, image deblurring is called blind and naturally presents a more difficult problem. This thesis focuses on two primary topics related to blind image deblurring. In the first part we work with the standard image deblurring based on the common convolution blur model and present a method of increasing robustness of the deblur- ring to phenomena violating the linear acquisition model, such as for example inten- sity clipping caused by sensor saturation in overexposed pixels. If not properly taken care of, these effects significantly decrease accuracy of the blur estimation and visual quality of the restored image. Rather than tailoring the deblurring method explicitly for each particular type of acquisition model violation we present a general approach based on flexible automatic...

This thesis deals with library for recording video in the background of 3D application. A library is designed to work under the Microsoft Windows and Linux operation systems.  It records image and also sound. Image recording is supported in OpenGL, Direct3D9, Direct3D10 and Direct3D11. To reduce video data size, library supports image compression using MJPG codec. Audio is recorded by WaveForm audio, Windows Core Audio or ALSA. Recorded sound is for whole operation system. A library is able to record up to two audio streams to accommodate possible microphone input. It can mix audio data together if needed. Output data are then written into AVI file. It is possible to write own text information as overlay that is rendered as part of application screen output.