1%% Reset everything
2
3clear all;
4clc;
5close all;
6addpath('helpers');
7
8%% Configure the benchmark
9
10% noncentral case
11cam_number = 4;
12% Getting 17 points, and testing all algorithms with the respective number of points
13pt_number = 17;
14% noise test, so no outliers
15outlier_fraction = 0.0;
16% repeat 1000 tests per noise level
17iterations = 1000;
18
19% The algorithms we want to test
20algorithms = { 'sixpt'; 'ge'; 'ge'; 'seventeenpt'; 'rel_nonlin_noncentral' };
21% This defines the number of points used for every algorithm
22indices = { [1:1:6]; [1:1:8]; [1:1:17]; [1:1:17]; [1:1:17] };
23% The name of the algorithms in the final plots
24names = { '6pt'; 'ge (8pt)'; 'ge (17pt)'; '17pt'; 'nonlin. opt. (17pt)' };
25
26% The maximum noise to analyze
27max_noise = 5.0;
28% The step in between different noise levels
29noise_step = 0.1;
30
31%% Run the benchmark
32
33%prepare the overall result arrays
34number_noise_levels = max_noise / noise_step + 1;
35num_algorithms = size(algorithms,1);
36mean_rotation_errors = zeros(num_algorithms,number_noise_levels);
37median_rotation_errors = zeros(num_algorithms,number_noise_levels);
38noise_levels = zeros(1,number_noise_levels);
39
40%Run the experiment
41for n=1:number_noise_levels
42
43    noise = (n - 1) * noise_step;
44    noise_levels(1,n) = noise;
45    display(['Analyzing noise level: ' num2str(noise)])
46
47    rotation_errors = zeros(num_algorithms,iterations);
48
49    counter = 0;
50
51    for i=1:iterations
52
53        % generate experiment
54        [v1,v2,t,R] = create2D2DOmniExperiment(pt_number,cam_number,noise,outlier_fraction);
55        [t_perturbed,R_perturbed] = perturb(t,R,0.01);
56        T_perturbed = [R_perturbed,t_perturbed];
57        T_init = [eye(3),zeros(3,1)];
58        T_gt = [R,t];
59
60        for a=1:num_algorithms
61
62            if strcmp(algorithms{a},'ge')
63                Out = opengv(algorithms{a},indices{a},v1,v2,T_init);
64            else
65                Out = opengv(algorithms{a},indices{a},v1,v2,T_perturbed);
66            end
67
68            if a > 3 %if a bigger than 4, we obtain entire transformation, and need to "cut" the rotation
69                temp = Out(:,1:3);
70                Out = temp;
71            end
72
73            rotation_error = evaluateRotationError( R, Out );
74            rotation_errors(a,i) = rotation_error;
75        end
76
77        counter = counter + 1;
78        if counter == 100
79            counter = 0;
80            display(['Iteration ' num2str(i) ' of ' num2str(iterations) '(noise level ' num2str(noise) ')']);
81        end
82    end
83
84    %Now compute the mean and median value of the error for each algorithm
85    for a=1:num_algorithms
86        mean_rotation_errors(a,n) = mean(rotation_errors(a,:));
87        median_rotation_errors(a,n) = median(rotation_errors(a,:));
88    end
89
90end
91
92%% Plot the results
93
94figure(1)
95plot(noise_levels,mean_rotation_errors,'LineWidth',2)
96legend(names,'Location','NorthWest')
97xlabel('noise level [pix]')
98ylabel('mean rot. error [rad]')
99grid on
100
101figure(2)
102plot(noise_levels,median_rotation_errors,'LineWidth',2)
103legend(names,'Location','NorthWest')
104xlabel('noise level [pix]')
105ylabel('median rot. error [rad]')
106grid on