|
|
Image Compression with Neural Networks
Teuer, Lukáš ; Sochor, Jakub (referee) ; Hradiš, Michal (advisor)
This document describes image compression using different types of neural networks. Features of neural networks like convolutional and recurrent networks are also discussed here. The document contains detailed description of various neural network architectures and their inner workings. In addition, experiments are carried out on various neural network structures and parameters in order to find the most appropriate properties for image compression. Also, there are proposed new concepts for image compression using neural networks that are also immediately tested. Finally, a network of the best concepts and parts discovered during experimentation is designed.
|
|
|
Analysis of JPEG 2000 Settings
Mitaš, Matěj ; Klíma, Ondřej (referee) ; Bařina, David (advisor)
This thesis analyses settings of JPEG 2000 format. The aim is to find appropriate values for each setting. Main criteria are compression performance, elapsed time and used memory. There are two libraries used that are implementing JPEG 2000 format: Kakadu and OpenJPEG. Testing was secured by framework implemented in Python. Results of this thesis enable users to use JPEG 2000 more effectively and with deeper understanding of inner workings.
|
|
|
Image Compression Using the Wavelet Transform
Přikryl, Lukáš ; Maršík, Lukáš (referee) ; Bařina, David (advisor)
The thesis discusses the image compression using the wavelet transform. The compression itself is done by EZW, SPIHT or EBCOT Tier1 algorithms. These algorithms are then compared to JPEG and JPEG2000 standards. Furthermore tiles, various wavelets and RGB and Y'CbCr color spaces were used to determine the influence on an image compression quality.
|
|
|
Video Player Based on FPGA
Sigmund, Stanislav ; Slaný, Karel (referee) ; Vašíček, Zdeněk (advisor)
This thesis deals with possible and realized decompression and playing of video on platforms, using FPGA unit. For implementation of this player is used platform FITKit, which has integrated VGA connector and large enough RAM memory. It uses a hard drive as memory medium with FAT32 file system.
|
|
|
Error Resilience Analysis for JPEG 2000
Kovalčík, Marek ; Klíma, Ondřej (referee) ; Bařina, David (advisor)
The aim of this thesis is to analyze modern image compression format of JPEG 2000. It analyzes the effect of error resilience mechanisms on image compression with different settings. The impact of using tag embedding to help repair damaged images or using compression modes to improve error resilience is examined here. Quality is evaluated by the PSNR metric that detects the similarity of compressed and reference file. Adding certain tags to the data stream or using certain compression modes should help secure the JPEG 2000 file against image reconstruction damage. To test this hipothesis, there was created a model that acidentally damage the compressed file and evaluate decompressed images. The Kakadu library, which provides efficient work with the JPEG 2000 format, is used for the work. The experimental data set consists of various photographs in uncompressed PPM format in smaller but also in higher resolutions. The result of this work is to find out which compression settings to use for which group of images to make the compression efficient and secure the best. The end of this thesis is devoted to comparison of error resilience of JPEG 2000 and CCSDS 122.0.
|
|
|
Analysis of JPEG 2000 Settings
Mitaš, Matěj ; Klíma, Ondřej (referee) ; Bařina, David (advisor)
This thesis analyses settings of JPEG 2000 format. The aim is to find appropriate values for each setting. Main criteria are compression performance, elapsed time and used memory. There are two libraries used that are implementing JPEG 2000 format: Kakadu and OpenJPEG. Testing was secured by framework implemented in Python. Results of this thesis enable users to use JPEG 2000 more effectively and with deeper understanding of inner workings.
|
|
|
Image Compression with Neural Networks
Teuer, Lukáš ; Sochor, Jakub (referee) ; Hradiš, Michal (advisor)
This document describes image compression using different types of neural networks. Features of neural networks like convolutional and recurrent networks are also discussed here. The document contains detailed description of various neural network architectures and their inner workings. In addition, experiments are carried out on various neural network structures and parameters in order to find the most appropriate properties for image compression. Also, there are proposed new concepts for image compression using neural networks that are also immediately tested. Finally, a network of the best concepts and parts discovered during experimentation is designed.
|
|
|
JPEG 2000 Implementation
Zlatohlávek, Adam ; Klíma, Ondřej (referee) ; Bařina, David (advisor)
The aim of this thesis is to propose a image compression methods in JPEG 2000. This consists of description of techniques used in base level stadard and analyze options in encoding process. The goal is to create compression process of input image from preprocessing to own output format. In the end of paperwork are presented results of own implementation in memory consuptions and encoding performance.
|
|
|
Automated Testing of Image Compressions
Matela, Vít ; Madzin, Matúš (referee) ; Luhan, Jan (advisor)
This thesis deals with the problem of automated testing software for image and video compression developed by the company Comprimato. This work was analyzed in the current testing environment in the company. Based on the resaults of this analysis, I designed and implemented a new automated testing system. Also included is a summary of the economic impact and recommendations for the future purchase of a new test hardware.
|
|
|
Lossy Image Compression
Bařina, David ; Herout, Adam (referee) ; Sumec, Stanislav (advisor)
This thesis deals with advanced methods of lossy image compression. To the midst of these methods belong the discrete cosine transform and Haar's discrete wavelet transform. Furthermore principles of color models and algorithms used in lossless data compression are explained in this text. The purpose of this thesis is the library implementation that is based on these cognizances.
|
guest ::
