|
|
Reconstruction of corrupted audio signals using deep unrolling
Kment, František ; Myška, Vojtěch (referee) ; Mokrý, Ondřej (advisor)
The thesis deals with the problem of audio signal restoration using traditional optimization methods combined with deep unrolling methods. An optimization task for filling in missing sections of the audio signal was formulated, and the proximal algorithm FISTA was chosen and subsequently implemented. Furthermore, three unrolled variants of the algorithm (Unrolled Fista Net) were implemented, two of which were further optimized using tests on a selected test dataset and trained on the Nsynth dataset. The results of the trained networks were compared both with competing methods and the original untrained variant of the algorithm. The comparison was made using objective metrics (MSE, SNR, PEAQ, PEMO-Q) and a subjective listening test.
|
|
|
Deinterlace Filter
Kuřina, Tomáš ; Španěl, Michal (referee) ; Beran, Vítězslav (advisor)
This document elaborates on the subject of video interlacing and its removal. It describes the interlacing of video, its history and the reasons that led to its use. The document also explains why it is necessary to remove interlacing and the basic methods that are used for it. It describes the proposed deinterlacing algorithm and its implementation, including description of inpainting and block matching. Included are also test results of both quality and speed of my deinterlacing algorithm. The final chapter describes the implementation as a console application and a DLL library.
|
|
|
Automatic Image Cut in Android
Ondriga, Patrik ; Uhříček, Daniel (referee) ; Kolář, Dušan (advisor)
Thesis is focused on suggest image cut. The idea was to create a mobile application that would suggest image cut from screenshot of a social network so after that, there was only image without social network environment. No AI was used for this purpose, but edge detection methods.
|
|
|
Restoration of degraded audiosignals using sparse representations
Mokrý, Ondřej ; Veselý, Vítězslav (referee) ; Rajmic, Pavel (advisor)
This bachelor thesis is focused on the problem of inpainting a segment of missing samples in an audiosignal. The signal is represented as sparse vector using discrete Gabor transform. The problem of inpainting missing samples while preserving the sparsity of the representation is formulated as an optimisation task, which is then solved using Douglas-Rachford algorithm. In contrast with the state-of-the-art approaches, the algorithm is extended by proposing method for compensating the energy decrease which occurs in the restored signal.
|
|
|
Applications of Dictionary Learning Methods for Audio Inpainting
Ozdobinski, Roman ; Rajmic, Pavel (referee) ; Mach, Václav (advisor)
This diploma thesis discusses methods of dictionary learning to inpaint missing sections in the audio signal. There was theoretically analyzed and practically used algorithms K-SVD and INK-SVD for dictionary learning. These dictionaries have been applied to the reconstruction of audio signals using OMP (Orthogonal Matching Pursuit). Furthermore, there was proposed an algorithm for selecting the stationary segments and their subsequent use as training data for K-SVD and INK-SVD. In the practical part of thesis have been observed efficiency with training set selection from whole signal compared with algorithm for stationary segmentation used. The influence of mutual coherence on the quality of reconstruction with incoherent dictionary was also studied. With created scripts for multiple testing in Matlab, there was performed comparison of these methods on genre distinct songs.
|
|
|
Image inpainting
Vodička, Radek ; Špiřík, Jan (referee) ; Rajmic, Pavel (advisor)
The aim of this paper is to describe the basic principles of inpainting methods. We take a closer look with method of continual correction for damaged area from its margins in poins and by means of level lines. Result of the work is to compare images where inpainting was applied, among images themseves as well as with original image,and evaluation which method is the best in the given situation and for the given type of images.
|
|
|
Modern methods for restoration of degraded audiosignals
Mokrý, Ondřej ; Koldovský,, Zbyněk (referee) ; Rajmic, Pavel (advisor)
The master's thesis deals with the problem of restoring a block of missing samples in a digital audio signal. This problem is formulated as an optimization task, which seeks the sparsest time-frequency representation of a signal within the set of feasible reconstructed signals. Several particular formulations are discussed, namely the analyzing and the synthesizing model, both for convex and non-convex approaches. Suitable algorithms are proposed for solving these formulations, and in the convex case, the method is further enhanced by various procedures to compensate for the energy drop in the inpainted signal segment. The proposed algorithms are tested on real recordings, and their performance is shown to be competitive with the state-of-the-art.
|
|
|
Image extrapolation methods
Ješko, Petr ; Špiřík, Jan (referee) ; Rajmic, Pavel (advisor)
The thesis deals with addition of pixels outside the image. Lists some methods for inpainting using computers and highlights the pitfalls that appear here. Examines methods for interpolation and approximation of functions in order to find the best method for extrapolating the image beyond its borders. Describes the basics of Wavelet transformation and Multiresolution analysis. It is proposed several methods for replenishment of pixels outside the image. PSNR and SSIM are used to compare achieved results. These methods are explained and compared. Briefly discusses the algorithm OMP, falling within the sparse representation of signals, and used in one of the methods. Also discussed is the development environment of MATLAB as a tool for the implementation of algorithms that practically solves the given problem. The practical part describes the implemented methods for adding pixels outside the image.
|
|
|
Image extrapolation methods
Ješko, Petr ; Špiřík, Jan (referee) ; Rajmic, Pavel (advisor)
The thesis deals with addition of pixels outside the image. Lists some methods for inpainting using computers and highlights the pitfalls that appear here. Examines methods for interpolation and approximation of functions in order to find the best method for extrapolating the image beyond its borders. Describes the basics of Wavelet transformation and Multiresolution analysis and briefly discusses about spatial filtering, edge detection and the algorithm OMP, falling within the sparse representation of signals. Theoretical knowledge of these areas are used in the design of several methods for adding pixels outside the image. PSNR and SSIM are used to compare achieved results. Also discussed is the development environment of MATLAB as a tool for the implementation of algorithms that practically solves the given problem.
|
|
|
Reconstruction and Enhancement of Damaged Parts of Fingerprint Images
Špila, Andrej ; Rydlo, Štěpán (referee) ; Heidari, Mona (advisor)
Táto práca sa zaoberá problémom rekonštrukcie snímkov odtlačkov prsta so zameraním na neobnovyteľné oblasti poškodené rôznymi kožnými ochoreniami. Generatívne súperi- ace siete s trénovateľnou konvolučnou vrstvou s gaborovými filtrami bola natrénovaná na dátovej sade reálnych snímkov odtlačkov prsta. Práca predvádza že natrénovaný model vie spoľahlivo rekonštruovať malé oblasti ľubovoľného tvaru a v prípade väčších oblastí, globálne skóre kvality rekonštruovaného odtlačku prsta získané využitím softvéru NIST biometric image software sa v porovnaní s originálnym snímkom navýšilo. Je navrhnutý štandardizovaný formát pre snímky odtlačkov prsta ktorý pomohol stabilizovat trénovanie generatívnych súperiacich sietí.
|
guest ::
