-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathload_3DOF_hyperparameters.m
More file actions
42 lines (32 loc) · 1.34 KB
/
Copy pathload_3DOF_hyperparameters.m
File metadata and controls
42 lines (32 loc) · 1.34 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
function [ OFeval_par ] = load_3DOF_hyperparameters()
% Hyperparameters used when optimizing the deformation vector field by grid search
%
% Author : Pohl Michel
% Date : Feb. 16th, 2021
% Version : v1.0
% License : 3-clause BSD License
% Threshold below which the image structure tensor is probably non invertible and the pseudo inverse is used for inversion.
OFeval_par.epsilon_detG = 0.001;
% Standard deviation of the Gaussian filter used for filtering the initial image
OFeval_par.sigma_init_tab = [0.2, 0.5, 1.0, 2.0];
% Standard deviation of the Gaussian filter used for filtering each layer of the pyramid
OFeval_par.sigma_subspl_tab = [0.2, 0.5, 1.0, 2.0];
% Choice of the method used to calculate the gradient
OFeval_par.grad_meth = 2;
% 2 : Scharr gradient
% 1 : Central difference
switch(OFeval_par.grad_meth)
case 1
OFeval_par.grad_meth_str = 'ctrl diff grdt';
case 2
OFeval_par.grad_meth_str = 'Schaar grdt';
end
% Standard deviation of the Gaussian kernel used in the Lucas Kanade method
OFeval_par.sigma_LK_tab = [1.0, 2.0, 3.0, 4.0];
% Number of layers in the pyramidal representation
OFeval_par.nb_layers_min = 1;
OFeval_par.nb_layers_max = 4;
% Number of iterations for refining the DVF at each layer
OFeval_par.nb_min_iter = 1;
OFeval_par.nb_max_iter = 3;
end