/
intensity_sample_ray_bundles.m
276 lines (245 loc) · 12.2 KB
/
intensity_sample_ray_bundles.m
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
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
function [] = intensity_sample_ray_bundles(im_filename, label, r_center, c_center)
%% setup
batch_mode = 1; % in batch mode, figures are invisible and no output is displayed
im.raw = imread(im_filename); % raw image filename
r_dim = size(im.raw, 1); % row dimension of the raw image
c_dim = size(im.raw, 2); % column dimension of the raw image
smooth_num = 100; % number of times to smooth the gray image
smooth_mask = [5 5]; % size of mask to smooth the gray image
smooth_std = 5.0; % standard deviation of the Gaussian function used to smooth the gray image
delta_theta_bundle = 2*pi; % angle of bundle: 2*pi (full circle => one bundle), pi/4 (1/8 circle => 8 bundles), etc.
delta_theta = pi/40; % angle between each extended sample ray: pi/40 => 80 rays per full circle, etc.
delta_rho = 1; % radial distance between intensity level samples (number of pixels)
delta_n = 0; % radius of neighborhood over which to average intensity levels at each sample
radius_mean_min_window = 1000; % maximum radial distance over which to compute the global minimum mean for determining the radius of each bundle
radius_median_min_window = 1000; % maximum radial distance over which to compute the global minimum median for determining the radius of each bundle
max_rho_sample = ceil(sqrt(r_dim^2+c_dim^2)/delta_rho); % maximum number of samples that each sample ray can take (for preallocating arrays)
max_theta_sample = ceil(2*pi/delta_theta); % maximum number of sample rays that can fit in a circle (for preallocating arrays, setting index limits)
max_theta_bundle = ceil(2*pi/delta_theta_bundle); % maximum number of bundles that can fit in a circle (for setting index limits)
bundle_samples = ceil(delta_theta_bundle/delta_theta); % number of sample rays that can fit in a bundle (for determining ranges)
center_num = numel(r_center); % number of centers to process
% create new results directory
path = strcat('/tmp/analysis/gradient/', label);
mkdir(path);
% open a data file
data_file = strcat(path, '/', 'data.txt');
fid = fopen(data_file, 'w');
fprintf(fid, 'number of bundles per circle = %f\n', max_theta_bundle);
fprintf(fid, 'number of sample rays per circle = %f\n', max_theta_sample);
fprintf(fid, '\n');
% convert RGB image to gray image
im.gray = rgb2gray(im.raw);
% smooth gray image
G = fspecial('gaussian', smooth_mask, smooth_std);
im.smooth = im.gray;
for i = 1 : smooth_num
im.smooth = imfilter(im.smooth, G, 'replicate', 'conv');
end
% for each center, measure intensity levels using bundles of sample rays, and plot the quantitative results for each bundle
for center_iter = 1 : center_num
fprintf(fid, 'center %d\n', center_iter);
% setup
r_init = r_center(center_iter);
c_init = c_center(center_iter);
measurements = 256 * ones(max_rho_sample, max_theta_sample);
% measure
for theta_iter = 1 : max_theta_sample
r = r_init;
c = c_init;
theta = (theta_iter - 1)*delta_theta;
rho = 0;
rho_iter = 1;
while r >= 1 && r <= r_dim && c >= 1 && c <= c_dim;
neighbor_val = [];
neighbor_num = 1;
for rr = -delta_n : delta_n
for cc = -delta_n : delta_n
if r+rr >= 1 && r+rr <= r_dim && c+cc >=1 && c+cc <= c_dim
neighbor_val(neighbor_num) = im.smooth(r+rr,c+cc);
neighbor_num = neighbor_num + 1;
end
end
end
val = mean(neighbor_val);
measurements(rho_iter,theta_iter) = val;
if ~batch_mode
fprintf('%d/%d: (%d,%f) = (%d,%d): %d\n', center_iter, center_num, rho, theta/pi, r, c, val);
end
rho = rho + delta_rho;
[x, y] = pol2cart(theta, rho);
r = r_init - floor(y);
c = c_init + floor(x);
rho_iter = rho_iter + 1;
end
end
% compute statistics and plot
for bundle_iter = 1 : max_theta_bundle
fprintf(fid, '\tbundle %d\n', bundle_iter);
measurements_bundle_range = (bundle_iter - 1)*bundle_samples + 1 : bundle_iter*bundle_samples;
measurements_mean = safe_stats(@mean, measurements(:,measurements_bundle_range)); % mean(measurements(:,measurements_bundle_range), 2);
measurements_std = safe_stats(@std, measurements(:,measurements_bundle_range)); % std(measurements(:,measurements_bundle_range), 0, 2);
measurements_cv = measurements_std ./ measurements_mean;
measurements_median = safe_stats(@median, measurements(:,measurements_bundle_range)); % median(measurements(:,measurements_bundle_range), 2);
measurements_median_min_locs = find_mins(measurements_median, radius_median_min_window);
measurements_radius = measurements_median_min_locs(1) * delta_rho;
measurements_range = 1:measurements_radius;
radii(center_iter,bundle_iter) = measurements_radius;
fprintf(fid, '\t\tradius = %d\n', measurements_radius);
if batch_mode
h = figure('Visible', 'off');
else
h = figure(bundle_iter);
end
clf;
subplot_width = 3;
% plot all trajectories, overlaid with mean and median
subplot(1,subplot_width,1);
hold on;
safe_plot(measurements(measurements_range,:));
plot(measurements_mean(measurements_range), 'b');
plot(measurements_median(measurements_range), 'r');
hold off;
title(['radius=', num2str(measurements_radius)]);
% plot the mean +/- std
fprintf(fid, '\t\tmean statistics\n');
subplot(1,subplot_width,2);
plot_time_series(measurements_mean', measurements_std', [0 0 1], 1, measurements_radius);
hold on;
[seg_i, seg_j] = segmented_least_squares(measurements_range, measurements_mean(measurements_range)', 200);
num_segs = numel(seg_i);
seg_a = zeros(1,num_segs);
seg_b = zeros(1,num_segs);
seg_e = zeros(1,num_segs);
for s = 1 : num_segs
fprintf(fid, '\t\t\tsegment %d\n', s);
[seg_a(s), seg_b(s)] = least_squares_fit(measurements_range, measurements_mean(measurements_range)', seg_i(s), seg_j(s));
seg_e(s) = least_squares_error(measurements_range, measurements_mean(measurements_range)', seg_i(s), seg_j(s));
plot(measurements_range(seg_i(s):seg_j(s)), seg_a(s) .* measurements_range(seg_i(s):seg_j(s)) + seg_b(s), 'k');
fprintf(fid, '\t\t\t\tbeg = %d\n', seg_i(s));
fprintf(fid, '\t\t\t\tend = %d\n', seg_j(s));
fprintf(fid, '\t\t\t\tlen = %d\n', seg_j(s) - seg_i(s) + 1);
fprintf(fid, '\t\t\t\tslo = %f\n', seg_a(s));
fprintf(fid, '\t\t\t\terr = %f\n', seg_e(s));
end
hold off;
title_foo = {};
for s = 1 : num_segs
title_foo{numel(title_foo)+1} = ['l=', num2str(seg_j(s) - seg_i(s) + 1), ', s=', sprintf('%0.2f', seg_a(s)), ', e=', sprintf('%0.2f', seg_e(s))];
end
title(title_foo);
% plot the median +/- std
fprintf(fid, '\t\tmedian statistics\n');
subplot(1,subplot_width,3);
plot_time_series(measurements_median', measurements_std', [1 0 0], 1, measurements_radius);
hold on;
[seg_i, seg_j] = segmented_least_squares(measurements_range, measurements_median(measurements_range)', 200);
num_segs = numel(seg_i);
seg_a = zeros(1,num_segs);
seg_b = zeros(1,num_segs);
seg_e = zeros(1,num_segs);
for s = 1 : numel(seg_i)
fprintf(fid, '\t\t\tsegment %d\n', s);
[seg_a(s), seg_b(s)] = least_squares_fit(measurements_range, measurements_median(measurements_range)', seg_i(s), seg_j(s));
seg_e(s) = least_squares_error(measurements_range, measurements_mean(measurements_range)', seg_i(s), seg_j(s));
plot(measurements_range(seg_i(s):seg_j(s)), seg_a(s) .* measurements_range(seg_i(s):seg_j(s)) + seg_b(s), 'k');
fprintf(fid, '\t\t\t\tbeg = %d\n', seg_i(s));
fprintf(fid, '\t\t\t\tend = %d\n', seg_j(s));
fprintf(fid, '\t\t\t\tlen = %d\n', seg_j(s) - seg_i(s) + 1);
fprintf(fid, '\t\t\t\tslo = %f\n', seg_a(s));
fprintf(fid, '\t\t\t\terr = %f\n', seg_e(s));
end
hold off;
title_foo = {};
for s = 1 : num_segs
title_foo{numel(title_foo)+1} = ['l=', num2str(seg_j(s) - seg_i(s) + 1), ', s=', sprintf('%0.2f', seg_a(s)), ', e=', sprintf('%0.2f', seg_e(s))];
end
title(title_foo);
fig_file = strcat(path, '/', 'blur_radius_', num2str(center_iter), '_bundle_', num2str(bundle_iter), '.pdf');
saveas(gcf, fig_file, 'pdf');
end
end
% plot the smoothed image, overlaid with centers, bundles, and bundle numbers
if batch_mode
h = figure('Visible', 'off');
else
h = figure(bundle_iter);
end
clf;
imshow(im.smooth);
hold on;
for center_iter = 1 : numel(r_center)
beg_x = [];
beg_y = [];
end_x = [];
end_y = [];
plot(c_center(center_iter), r_center(center_iter), 'ro');
text(c_center(center_iter) + 30, r_center(center_iter), num2str(center_iter), 'Color', 'r');
for bundle_iter = 1 : max_theta_bundle
theta_range = (bundle_iter - 1)*delta_theta_bundle : 0.01 : bundle_iter*delta_theta_bundle;
[perim_x, perim_y] = pol2cart(theta_range, radii(center_iter,bundle_iter));
plot(c_center(center_iter) + perim_x, r_center(center_iter) - perim_y, 'r');
[label_x, label_y] = pol2cart(theta_range(floor(numel(theta_range)/2)), radii(center_iter,bundle_iter) + 30);
text(c_center(center_iter) + label_x, r_center(center_iter) - label_y, num2str(bundle_iter), 'Color', 'r');
beg_x(bundle_iter) = c_center(center_iter) + perim_x(1);
beg_y(bundle_iter) = r_center(center_iter) - perim_y(1);
end_x(bundle_iter) = c_center(center_iter) + perim_x(numel(perim_x));
end_y(bundle_iter) = r_center(center_iter) - perim_y(numel(perim_y));
end
for bundle_iter = 1 : max_theta_bundle - 1
plot([end_x(bundle_iter) beg_x(bundle_iter+1)], [end_y(bundle_iter) beg_y(bundle_iter+1)], 'r');
end
plot([end_x(bundle_iter+1) beg_x(1)], [end_y(bundle_iter+1) beg_y(1)], 'r');
end
hold off;
fig_file = strcat(path, '/', 'blur_radii.pdf');
saveas(gcf, fig_file, 'pdf');
% close the data file
fclose(fid);
return;
%% plot a time series average +/- standard deviation (average curve surrounded by +/- gray patches)
% a: average (m time series x n ticks)
% s: standard deviation (m time series x n ticks)
% c: color map (m time series x 3 {r,g,b})
% b: begin time (integer)
% e: end time (integer)
function [] = plot_time_series(a, s, c, b, e)
domain = b : e;
gray = [0.9 0.9 0.9];
hold on;
for t = 1 : size(a,1)
patch([domain fliplr(domain)], [a(t,domain) - s(t,domain), fliplr(a(t,domain) + s(t,domain))], gray, 'LineStyle', 'none');
plot(domain, a(t,domain), 'color', c(t,:));
end
hold off;
end % function plot_time_series
function [l] = find_mins(a, w)
[m, l] = min(a(1:w));
end
function [F] = safe_stats(f, M)
num_r = size(M,1);
num_c = size(M,2);
F = [];
for r = 1 : num_r
safe_set = [];
for c = 1 : num_c
if M(r,c) ~= 256
safe_set(numel(safe_set)+1) = M(r,c);
end
end
F(r,1) = f(safe_set);
end
end
function [] = safe_plot(M)
num_r = size(M,1);
num_c = size(M,2);
for c = 1 : num_c
for r = 1 : num_r
if M(r,c) == 256
break;
end
end
safe_r_idx = r - 1;
plot(1:safe_r_idx, M(1:safe_r_idx, c), 'Color', [0.9 0.9 0.9]);
end
end
end