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
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
|
%% Reset everything
clear all;
clc;
close all;
addpath('helpers');
%% Configure the benchmark
% noncentral case
cam_number = 4;
% let's only get 6 points, and generate new ones in each iteration
pt_number = 50;
% noise test, so no outliers
outlier_fraction = 0.0;
% repeat 5000 iterations per noise level
iterations = 5000;
% The algorithms we want to test
algorithms = { 'gp3p'; 'gpnp'; 'gpnp'; 'abs_nonlin_noncentral'; 'abs_nonlin_noncentral'; 'upnp'; 'upnp' };
% This defines the number of points used for every algorithm
indices = { [1, 2, 3]; [1:1:10]; [1:1:50]; [1:1:10]; [1:1:50]; [1:1:10]; [1:1:50] };
% The name of the algorithms on the plots
names = { 'GP3P'; 'GPnP (10pts)'; 'GPnP (50pts)'; 'nonlin. opt. (10pts)'; 'nonlin. opt. (50pts)'; 'UPnP (10pts)'; 'UPnP (50pts)' };
% The maximum noise to analyze
max_noise = 5.0;
% The step in between different noise levels
noise_step = 0.1;
%% Run the benchmark
%prepare the overall result arrays
number_noise_levels = max_noise / noise_step + 1;
num_algorithms = size(algorithms,1);
mean_position_errors = zeros(num_algorithms,number_noise_levels);
mean_rotation_errors = zeros(num_algorithms,number_noise_levels);
median_position_errors = zeros(num_algorithms,number_noise_levels);
median_rotation_errors = zeros(num_algorithms,number_noise_levels);
noise_levels = zeros(1,number_noise_levels);
%Run the experiment
for n=1:number_noise_levels
noise = (n - 1) * noise_step;
noise_levels(1,n) = noise;
display(['Analyzing noise level: ' num2str(noise)])
position_errors = zeros(num_algorithms,iterations);
rotation_errors = zeros(num_algorithms,iterations);
counter = 0;
validIterations = 0;
for i=1:iterations
% generate experiment
[points,v,t,R] = create2D3DExperiment(pt_number,cam_number,noise,outlier_fraction);
[t_perturbed,R_perturbed] = perturb(t,R,0.01);
T_perturbed = [R_perturbed,t_perturbed];
T_gt = [R,t];
% run all algorithms
allValid = 1;
for a=1:num_algorithms
T = opengv(algorithms{a},indices{a},points,v,T_perturbed);
[position_error, rotation_error] = evaluateTransformationError( T_gt, T );
if( position_error > 100 )
allValid = 0;
break;
else
position_errors(a,validIterations+1) = position_error;
rotation_errors(a,validIterations+1) = rotation_error;
end
end
if allValid == 1
validIterations = validIterations +1;
end
counter = counter + 1;
if counter == 100
counter = 0;
display(['Iteration ' num2str(i) ' of ' num2str(iterations) '(noise level ' num2str(noise) ')']);
end
end
%Now compute the mean and median value of the error for each algorithm
for a=1:num_algorithms
mean_position_errors(a,n) = mean(position_errors(a,1:validIterations));
median_position_errors(a,n) = median(position_errors(a,1:validIterations));
mean_rotation_errors(a,n) = mean(rotation_errors(a,1:validIterations));
median_rotation_errors(a,n) = median(rotation_errors(a,1:validIterations));
end
end
%% Plot the results
figure(1)
plot(noise_levels',mean_rotation_errors','LineWidth',2)
legend(names,'Location','NorthWest')
xlabel('noise level [pix]')
ylabel('mean rot. error [rad]')
grid on
figure(2)
plot(noise_levels',median_rotation_errors','LineWidth',2)
legend(names,'Location','NorthWest')
xlabel('noise level [pix]')
ylabel('median rot. error [rad]')
grid on
figure(3)
plot(noise_levels',mean_position_errors','LineWidth',2)
legend(names,'Location','NorthWest')
xlabel('noise level [pix]')
ylabel('mean pos. error [m]')
grid on
figure(4)
plot(noise_levels',median_position_errors','LineWidth',2)
legend(names,'Location','NorthWest')
xlabel('noise level [pix]')
ylabel('median pos. error [m]')
grid on
|
