|
|
Sparse Signal Recovery via Convex Optimization
Novosadová, Michaela
We propose recovering 1D piecewice linear signal using a sparsity-based method consisting of two steps. The first step is signal segmentation via optimization algorithms solving sparsity based model. Second step consists of applying an ordinary mean square method on each detected segment of the signal. We show results of our experiments on two types of the signal.
|
|
| |
|
|
Engineering Optimization
Kokrda, Lukáš ; Hrabec, Dušan (referee) ; Popela, Pavel (advisor)
The bachelor thesis deals with convex optimization and in particular, it processes a design of the optimal support of a loaded beam. For better understanding of the terms used, the bachelor thesis contains the brief introduction to the convex optimization problems, explanation of the basic therms of ordinary dierential equations and theory of elasticity. When the original model is built, then the results are obtained by computations in the MATLAB software.
|
|
|
Restoration of overexposed digital image
Zonygová, Kristýna ; Veselý, Vítězslav (referee) ; Rajmic, Pavel (advisor)
The Bachelor thesis deals with the recovery of overexposed grayscale images based on sparse signal representation. For image conversion to another representation wavelet transform was used. In this representation, the solution of convex optimization which demand on relax signal sparse and restored pixels values was searched for. In this case, the proximal Douglas-Rachford algorithm was applied which uses two proximal operators. The implementation was carried out in numerical computing enviroment MATLAB using the Wavelet Toolbox software. The PSNR (peak signal-to-noise ratio) was utilized to evaluate success rate of proposed method. The method was tested on 5 random images and compared to results of image manipulation software Adobe Photoshop Lightroom CC.
|
|
| |
|
|
Restoration of overexposed digital image
Zonygová, Kristýna ; Veselý, Vítězslav (referee) ; Rajmic, Pavel (advisor)
The Bachelor thesis deals with the recovery of overexposed grayscale images based on sparse signal representation. For image conversion to another representation wavelet transform was used. In this representation, the solution of convex optimization which demand on relax signal sparse and restored pixels values was searched for. In this case, the proximal Douglas-Rachford algorithm was applied which uses two proximal operators. The implementation was carried out in numerical computing enviroment MATLAB using the Wavelet Toolbox software. The PSNR (peak signal-to-noise ratio) was utilized to evaluate success rate of proposed method. The method was tested on 5 random images and compared to results of image manipulation software Adobe Photoshop Lightroom CC.
|
|
|
Sparse Signal Recovery via Convex Optimization
Novosadová, Michaela
We propose recovering 1D piecewice linear signal using a sparsity-based method consisting of two steps. The first step is signal segmentation via optimization algorithms solving sparsity based model. Second step consists of applying an ordinary mean square method on each detected segment of the signal. We show results of our experiments on two types of the signal.
|
|
|
Engineering Optimization
Kokrda, Lukáš ; Hrabec, Dušan (referee) ; Popela, Pavel (advisor)
The bachelor thesis deals with convex optimization and in particular, it processes a design of the optimal support of a loaded beam. For better understanding of the terms used, the bachelor thesis contains the brief introduction to the convex optimization problems, explanation of the basic therms of ordinary dierential equations and theory of elasticity. When the original model is built, then the results are obtained by computations in the MATLAB software.
|
guest ::
