Class: Panomosity::Panorama

Inherits:
Object
  • Object
show all
Includes:
Utils
Defined in:
lib/panomosity/panorama.rb

Instance Attribute Summary collapse

Instance Method Summary collapse

Methods included from Utils

#calculate_average, #calculate_average_and_std, #remove_outliers

Constructor Details

#initialize(input, options = {}) ⇒ Panorama

Returns a new instance of Panorama.



9
10
11
12
13
14
15
16
17
18
19
 
# File 'lib/panomosity/panorama.rb', line 9

def initialize(input, options = {})
  @input = input
  @options = options
  @options[:verbosity] ||= 0
  @images = Image.parse(@input)
  @variable = PanoramaVariable.parse(@input).first
  ControlPoint.parse(@input)
  @control_points = ControlPoint.calculate_distances(@images, @variable)
  @optimisation_variables = OptimisationVariable.parse(@input)
  @logger = Panomosity.logger
end

Instance Attribute Details

#control_points ⇒ Object

Returns the value of attribute control_points.



7
8
9
 
# File 'lib/panomosity/panorama.rb', line 7

def control_points
  @control_points
end

#images ⇒ Object

Returns the value of attribute images.



7
8
9
 
# File 'lib/panomosity/panorama.rb', line 7

def images
  @images
end

#logger ⇒ Object

Returns the value of attribute logger.



7
8
9
 
# File 'lib/panomosity/panorama.rb', line 7

def logger
  @logger
end

#optimisation_variables ⇒ Object

Returns the value of attribute optimisation_variables.



7
8
9
 
# File 'lib/panomosity/panorama.rb', line 7

def optimisation_variables
  @optimisation_variables
end

#options ⇒ Object

Returns the value of attribute options.



7
8
9
 
# File 'lib/panomosity/panorama.rb', line 7

def options
  @options
end

#variable ⇒ Object

Returns the value of attribute variable.



7
8
9
 
# File 'lib/panomosity/panorama.rb', line 7

def variable
  @variable
end

Instance Method Details

#attributes ⇒ Object 



352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
 
# File 'lib/panomosity/panorama.rb', line 352

def attributes
  calculate_neighborhoods
  control_points = self.control_points.dup
  control_points.each_with_index { |cp, i| cp[:id] = i }
  neighborhoods = GeneralizedNeighborhood.neighborhoods.dup
  neighborhoods.each_with_index { |n, i| n.id = i }
  types = %i(horizontal vertical)
  similar_neighborhoods = types.map { |type| GeneralizedNeighborhood.similar_neighborhoods(type: type) }.flatten
  neighborhoods_by_similar_neighborhood = types.map { |type| GeneralizedNeighborhood.neighborhoods_by_similar_neighborhood(type: type) }.flatten
  {
    images: images.map(&:attributes),
    variable: variable.attributes,
    control_points: control_points.map(&:attributes),
    optimisation_variables: optimisation_variables.map(&:attributes),
    pairs: Pair.all.map(&:attributes),
    similar_neighborhoods: similar_neighborhoods.map(&:attributes),
    neighborhoods_by_similar_neighborhood: neighborhoods_by_similar_neighborhood.map(&:attributes)
  }
end

#calculate_neighborhoods ⇒ Object 



21
22
23
 
# File 'lib/panomosity/panorama.rb', line 21

def calculate_neighborhoods
  GeneralizedNeighborhood.calculate_all(panorama: self, options: options)
end

#calibration? ⇒ Boolean

Returns:

  • (Boolean) 


336
337
338
 
# File 'lib/panomosity/panorama.rb', line 336

def calibration?
  !!@input.split(/\n/).find { |line| line == '#panomosity calibration true' }
end

#clean_control_points ⇒ Object 



25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
 
# File 'lib/panomosity/panorama.rb', line 25

def clean_control_points
  options.merge!(distances: { x1: 30, x2: 30 }) if calibration? && !options[:distances].nil?
  options.merge!(regional_distance_similarities_count: 2) unless options[:regional_distance_similarities_count]
  calculate_neighborhoods

  control_points_to_keep = Pair.select_control_points_with_regional_distance_similarities
  bad_control_points = control_points.reject { |cp| control_points_to_keep.map(&:raw).include?(cp.raw) }
  # far_control_points = control_points.select { |cp| cp.prdist > 50 }
  control_points_to_clean = bad_control_points.uniq(&:raw)

  # log warnings
  control_point_ratio = control_points_to_clean.count.to_f / control_points.count
  logger.warn "Removing more than 30% (#{(control_point_ratio * 100).round(4)}%) of control points. May potentially cause issues." if control_point_ratio >= 0.3
  control_point_pair_ratio = Pair.without_enough_control_points(ignore_connected: true).count.to_f / Pair.all.count
  logger.warn "More than 50% (#{(control_point_pair_ratio * 100).round(4)}%) of pairs have fewer than 3 control points. May potentially cause issues." if control_point_pair_ratio >= 0.5

  control_points_to_clean
end

#create_calibration_report ⇒ Object 



304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
 
# File 'lib/panomosity/panorama.rb', line 304

def create_calibration_report
  # create a file if one doesn't exist
  filename = 'calibration_report.json'
  unless File.file?(filename)
    logger.info 'creating calibration_report.json since one does not exist'
    File.open(filename, 'w+') { |f| f.puts '{}' }
  end

  calibration_report = JSON.parse(File.read(filename))

  if @options[:report_type] == 'position'
    options.merge!(distances: { x1: 30, x2: 30 }) if calibration? && !options[:distances].nil?
    calculate_neighborhoods

    xh_avg = GeneralizedNeighborhood.horizontal.first.x_avg
    yh_avg = GeneralizedNeighborhood.horizontal.first.y_avg
    xv_avg = GeneralizedNeighborhood.vertical.first.x_avg
    yv_avg = GeneralizedNeighborhood.vertical.first.y_avg
    calibration_report['position'] = {
      xh_avg: xh_avg,
      yh_avg: yh_avg,
      xv_avg: xv_avg,
      yv_avg: yv_avg
    }
  else
    calibration_report['roll'] = images.first.r
  end

  logger.info 'writing calibration_report.json'
  File.open(filename, 'w+') { |f| f.puts calibration_report.to_json }
end

#diagnose ⇒ Object 



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
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
 
# File 'lib/panomosity/panorama.rb', line 154

def diagnose
  calculate_neighborhoods

  recommendations = []
  messages = []

  logger.debug "total number of control points: #{control_points.count}"
  logger.debug "total number of generated control points: #{control_points.select(&:generated?).count}"
  logger.debug "total number of not generated control points: #{control_points.select(&:not_generated?).count}"

  control_point_pair_ratio = Pair.without_enough_control_points(ignore_connected: true).count.to_f / Pair.all.count
  if control_point_pair_ratio >= 0.5
    message = <<~MESSAGE
      More than 50% (#{(control_point_pair_ratio*100).round(4)}%) of pairs have fewer than 3 control points.
      May potentially cause issues.
    MESSAGE
    logger.warn message
    messages << message
  end

  control_point_generated_ratio = control_points.select(&:generated?).count.to_f / control_points.select(&:not_generated?).count
  if control_point_generated_ratio >= 0.3
    message = <<~MESSAGE
      More than 30% (#{(control_point_generated_ratio*100).round(4)}%) control points were generated.
      This indicates a failure to find control points between images pairs due to poor lighting or insufficient complexity.
    MESSAGE
    logger.warn message
    messages << message
  end

  # neighborhood group tests
  group_count = GeneralizedNeighborhood.horizontal.count
  if group_count < 5
    message = <<~MESSAGE
      Total number of horizontal neighborhood groups is #{group_count} which is very low.
      This can mean either low variation in control points distances or that not enough control points could be found.
    MESSAGE
    logger.warn message
    messages << message
  end

  group_std_avg = calculate_average(values: GeneralizedNeighborhood.horizontal[0..4].map(&:dist_std))
  if group_std_avg > 1.0
    message = <<~MESSAGE
      The standard deviation of distances in the top 5 horizontal neighborhood groups is #{group_std_avg} which is high.
      The standard deviation implies that control points neighborhoods making up this group can vary more than 1.0 in distance.
      On highly optimized images (with many good control points) this standard deviation should be near 0.
      This could mean that even after optimization, there may be a seam on an individual pair.
      This also means that the images may represent a 3D object that has perspective differences.
    MESSAGE
    logger.warn message
    messages << message
  end

  group_control_points = GeneralizedNeighborhood.horizontal.first.control_points.count
  total_control_points = Pair.horizontal.map(&:control_points).flatten.uniq(&:raw).count
  group_control_point_ratio = group_control_points.to_f / total_control_points
  if group_control_point_ratio < 0.2
    message = <<~MESSAGE
      Less than 20% (#{(group_control_point_ratio*100).round(4)}%) of horizontal control points in the best
      horizontal neighborhood group (#{group_control_points}) make up the total number of horizontal control points (#{total_control_points}).
      This means panosmosity failed to find a neighborhood group that would include enough similarities between control point distances.
      There will very likely be seams horizontally.
    MESSAGE
    logger.warn message
    messages << message
    recommendations << 'horizontal'
  end

  group_count = GeneralizedNeighborhood.vertical.count
  if group_count < 5
    message = <<~MESSAGE
      Total number of vertical neighborhood groups is #{group_count} which is very low.
      This can mean either low variation in control points distances or that not enough control points could be found.
    MESSAGE
    logger.warn message
    messages << message
  end

  group_std_avg = calculate_average(values: GeneralizedNeighborhood.vertical[0..4].map(&:dist_std))
  if group_std_avg > 1.0
    message = <<~MESSAGE
      The standard deviation of distances in the top 5 vertical neighborhood groups is #{group_std_avg} which is high.
      The standard deviation implies that control points neighborhoods making up this group can vary more than 1.0 in distance.
      On highly optimized images (with many good control points) this standard deviation should be near 0.
      This could mean that even after optimization, there may be a seam on an individual pair.
      This also means that the images may represent a 3D object that has perspective differences.
    MESSAGE
    logger.warn message
    messages << message
  end

  group_control_points = GeneralizedNeighborhood.vertical.first.control_points.count
  total_control_points = Pair.vertical.map(&:control_points).flatten.uniq(&:raw).count
  group_control_point_ratio = group_control_points.to_f / total_control_points
  if group_control_point_ratio < 0.2
    message = <<~MESSAGE
      Less than 20% (#{(group_control_point_ratio*100).round(4)}%) of vertical control points in the best
      vertical neighborhood group (#{group_control_points}) make up the total number of vertical control points (#{total_control_points}).
      This means panosmosity failed to find a neighborhood group that would include enough similarities between control point distances.
      There will very likely be seams vertically.
    MESSAGE
    logger.warn message
    recommendations << 'vertical'
  end

  logger.info 'creating diagnostic_report.json'

  pair = Pair.horizontal.first
  delta_d = pair.first_image.d - pair.last_image.d
  roll = pair.first_image.r
  pair = Pair.vertical.first
  delta_e = pair.first_image.e - pair.last_image.e

  diagnostic_report = {
    messages: messages,
    recommendations: recommendations,
    data: {
      delta_d: delta_d,
      delta_e: delta_e,
      roll: roll,
      horizontal: GeneralizedNeighborhood.horizontal.first.attributes,
      vertical: GeneralizedNeighborhood.vertical.first.attributes
    }
  }
rescue => error
  message = "Got error #{error.message} when calculating neighborhoods. Recommending fallback"

  logger.error message
  error.backtrace.each { |line| logger.error line }

  recommendations = %w(horizontal vertical)

  diagnostic_report = {
    messages: messages,
    recommendations: recommendations,
    data: {}
  }
ensure
  File.open('diagnostic_report.json', 'w+') { |f| f.puts diagnostic_report.to_json }

  if recommendations.empty?
    logger.warn 'No recommendations'
    puts 'none'
  else
    logger.warn 'Recommendations are to regenerate with control points generated from calibration cards:'
    puts recommendations.join(',')
  end
end

#fix_unconnected_image_pairs ⇒ Object 



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
 
# File 'lib/panomosity/panorama.rb', line 44

def fix_unconnected_image_pairs
  logger.info 'finding unconnected image pairs'

  options.merge!(distances: { x1: 30, x2: 30 }) if calibration? && !options[:distances].nil?
  calculate_neighborhoods

  unconnected_image_pairs = Pair.unconnected
  logger.debug unconnected_image_pairs.map { |i| { type: i.type, pair: i.pair.map(&:id) } }

  logger.info 'finding control points with unrealistic distances (<1)'
  bad_control_points = control_points.select { |cp| cp.pdist <= 1.0 }
  logger.info 'adding pairs that have do not have enough control points (<3)'
  changing_control_points_pairs = Pair.without_enough_control_points
  changed_pairs = []

  logger.info 'writing new control points'
  control_point_lines_started = false
  @lines = @input.each_line.map do |line|
    cp = ControlPoint.parse_line(line)
    if cp.nil?
      # Control point lines ended
      if control_point_lines_started
        control_point_lines_started = false
        unconnected_image_pairs.map do |pair|
          generate_control_points(pair: pair, message: 'adding control points connecting')
        end + [line]
      else
        next line
      end
    else
      control_point_lines_started = true
      bad_control_point = bad_control_points.find { |c| c.raw == line }
      changing_control_point_pair = changing_control_points_pairs.find { |pair| pair.control_points.find { |c| c.raw == line } }

      if bad_control_point
        generate_control_points(bad_control_point: bad_control_point, message: 'replacing unrealistic control point connecting')
      elsif changing_control_point_pair && !changed_pairs.include?(changing_control_point_pair.to_s)
        changed_pairs << changing_control_point_pair.to_s
        bad_control_point = changing_control_point_pair.control_points.first
        generate_control_points(bad_control_point: bad_control_point, message: 'adding control points connecting')
      else
        next line
      end
    end
  end.compact.flatten
end

#generate_control_points(pair: nil, bad_control_point: nil, message: '') ⇒ Object 



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
 
# File 'lib/panomosity/panorama.rb', line 91

def generate_control_points(pair: nil, bad_control_point: nil, message: '')
  if pair
    if pair.horizontal?
      group = GeneralizedNeighborhood.horizontal.first
    else
      group = GeneralizedNeighborhood.vertical.first
    end
  else
    if bad_control_point.conn_type == :horizontal
      group = GeneralizedNeighborhood.horizontal.first
    else
      group = GeneralizedNeighborhood.vertical.first
    end
  end

  control_point = ControlPoint.new(group.center.center.attributes(raw: true))

  if pair
    control_point[:n] = pair.first_image.id
    control_point[:N] = pair.last_image.id
  else
    control_point[:n] = bad_control_point[:n]
    control_point[:N] = bad_control_point[:N]
  end

  image_1 = images.find { |i| i.id == control_point[:n] }
  image_2 = images.find { |i| i.id == control_point[:N] }

  x_diff = group.x_avg + (image_2.d - image_1.d)
  y_diff = group.y_avg + (image_2.e - image_1.e)

  x1 = x_diff <= 0 ? -x_diff + 15 : 0
  y1 = y_diff <= 0 ? -y_diff + 15 : 0

  control_point[:x] = x1
  control_point[:X] = x1 + x_diff
  control_point[:y] = y1
  control_point[:Y] = y1 + y_diff

  logger.info "#{message} #{control_point.n1} <> #{control_point.n2}"
  i = images.first
  3.times.map do
    if control_point.conn_type == :horizontal
      control_point[:y] += i.h * 0.25
      control_point[:Y] += i.h * 0.25
    else
      control_point[:x] += i.w * 0.25
      control_point[:X] += i.w * 0.25
    end
    # marks the control point as generated

    control_point[:g] = 0
    control_point.to_s
  end.join
end

#get_neighborhood_info ⇒ Object 



147
148
149
150
151
152
 
# File 'lib/panomosity/panorama.rb', line 147

def get_neighborhood_info
  Pair.calculate_neighborhoods(self)
  Pair.calculate_neighborhood_groups
  Pair.info
  NeighborhoodGroup.info
end

#save_file(filename) ⇒ Object 



340
341
342
343
344
345
346
347
348
349
350
 
# File 'lib/panomosity/panorama.rb', line 340

def save_file(filename)
  logger.info "saving file #{filename}"

  lines = @input.each_line.map do |line|
    objects = [images, variable, control_points, optimisation_variables].flatten
    object = objects.find { |object| object.raw == line }
    object&.to_s || line
  end.compact

  File.open(filename, 'w') { |f| lines.each { |line| f.puts line } }
end