sqed 0.1.5 → 0.1.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 11687929d5850b325870f2f76a4d628f2707f324
-  data.tar.gz: f38950c622a96cda352855e494e97ce3bdb8bf8e
+  metadata.gz: bd0958fc2efbed976b77385f511ce9025f26b5b2
+  data.tar.gz: 2f3efab45b172677057170dfdbdee8366372fac0
 SHA512:
-  metadata.gz: 486844c8d1499848dbab9bedfd69155a023eac59d2470914cdec6d80c58a8457db398386ada6f9d4b153e549ece076c471279d3ef768be13333bb767f78d1e9e
-  data.tar.gz: bcf06a0b5d27545bbac8123a1dd69f84166f37d760dff9a95e33b5e8d73f688ed8d12cd419ec84a76011fd328c4aa1ccaa58ecaed863a94a22fd2d5ab82fc424
+  metadata.gz: 04c09b12b5212a5b7c6cf356caa6e04ef4c422b5e83f14c652996089b9f30c016102572cc2963c2710d56135d7411e2778303034de9b8a44597d4feee7796b43
+  data.tar.gz: 14a8a7cd8bd19a6c3c00e3105b82b513f119bca6f3febac4751e9d1714e2ff08d4aaa3a5bbd3ead4f349d56078830aba02272bc3231424cea3877064b5b9f5b4

data/lib/sqed.rb

@@ -41,7 +41,10 @@ class Sqed
   # a symbol, :red, :green, :blue, describing the boundary color within the stage 
   attr_accessor :boundary_color
 
-  def initialize(image: image, pattern: pattern, has_border: true, boundary_color: :green)
+  # Boolean, whether to do the boundary detection (not stage detection at present) against a thumbnail version of the passed image (faster, less accurate, true be default) 
+  attr_accessor :use_thumbnail
+
+  def initialize(image: image, pattern: pattern, has_border: true, boundary_color: :green, use_thumbnail: true)
     raise 'extraction pattern not defined' if pattern && !SqedConfig::EXTRACTION_PATTERNS.keys.include?(pattern) 
  
     @image = image
@@ -51,7 +54,7 @@ class Sqed
     @pattern = pattern
     @pattern ||= :cross
     @boundary_color = boundary_color
-
+    @use_thumbnail = use_thumbnail
     set_stage_boundary if @image
   end
 
@@ -108,6 +111,7 @@ class Sqed
     extractor.result
   end
 
+  # Debugging purposes
   def attributes
     {
       image: @image,
@@ -115,7 +119,8 @@ class Sqed
       stage_boundary: stage_boundary,
       has_border: @has_border,
       pattern: @pattern,
-      boundary_color: @boundary_color
+      boundary_color: @boundary_color, 
+      use_thumbnail: @use_thumbnail
     }
   end
 
@@ -138,11 +143,12 @@ class Sqed
   def get_section_boundaries
     boundary_finder_class = SqedConfig::EXTRACTION_PATTERNS[@pattern][:boundary_finder]
 
-    options = {image: stage_image}
+    options = {image: stage_image, use_thumbnail: use_thumbnail}
     options.merge!( layout: SqedConfig::EXTRACTION_PATTERNS[@pattern][:layout] ) unless  boundary_finder_class.name == 'Sqed::BoundaryFinder::CrossFinder'
-    options.merge!( boundary_color: @boundary_color) if  boundary_finder_class.name == 'Sqed::BoundaryFinder::ColorLineFinder'
+    options.merge!( boundary_color: @boundary_color) if boundary_finder_class.name == 'Sqed::BoundaryFinder::ColorLineFinder'
 
     boundary_finder_class.new(options).boundaries
+
   end
 
 end

data/lib/sqed/boundaries.rb

@@ -91,4 +91,18 @@ class Sqed::Boundaries
     end
     true 
   end
+
+  def zoom(width_factor, height_factor)
+    coordinates.keys.each do |i|
+      set(i, [   
+        (x_for(i).to_f * width_factor).to_i,
+        (y_for(i).to_f * height_factor).to_i,
+        (width_for(i).to_f * width_factor).to_i,
+        (height_for(i).to_f * height_factor).to_i
+      ])
+    
+    end  
+  end
+
+
 end

data/lib/sqed/boundary_finder.rb

@@ -2,6 +2,9 @@
 # return derivative images. Finders operate on cropped images, i.e. only the "stage".
 #
 class Sqed::BoundaryFinder
+
+  THUMB_SIZE = 100 
+
   # the passed image
   attr_reader :img
 
@@ -11,10 +14,18 @@ class Sqed::BoundaryFinder
   # A Sqed::Boundaries instance, stores the coordinates of all of the layout sections 
   attr_reader :boundaries
 
-  def initialize(image: image, layout: layout)
+  # Whether to compress the original image to a thumbnail when finding boundaries
+  attr_reader :use_thumbnail
+
+  # when we compute using a derived thumbnail we temporarily store the full size image here
+  attr_reader :original_image
+
+  def initialize(image: image, layout: layout, use_thumbnail: true)
     raise 'No layout provided.' if layout.nil?
     raise 'No image provided.' if image.nil? || image.class.name != 'Magick::Image'
 
+    @use_thumbnail = use_thumbnail
+
     @layout = layout
     @img = image
     true
@@ -25,12 +36,46 @@ class Sqed::BoundaryFinder
     @boundaries ||= Sqed::Boundaries.new(@layout)
   end
 
+  def longest_thumbnail_axis
+    img.columns > img.rows ? :width : :height 
+  end
+
+  def thumbnail_height
+    if longest_thumbnail_axis == :height
+      THUMB_SIZE
+    else
+      (img.rows.to_f * (THUMB_SIZE.to_f / img.columns.to_f)).round.to_i
+    end
+  end
+
+  def thumbnail_width
+    if longest_thumbnail_axis == :width
+      THUMB_SIZE
+    else
+      (img.columns.to_f * (THUMB_SIZE.to_f / img.rows.to_f)).round.to_i
+    end
+  end
+
+  # see https://rmagick.github.io/image3.html#thumbnail
+  def thumbnail
+    img.thumbnail(thumbnail_width, thumbnail_height)
+  end
+
+  def width_factor
+    img.columns.to_f / thumbnail_width.to_f
+  end
+
+  def height_factor
+    img.rows.to_f / thumbnail_height.to_f
+  end
+
+  def zoom_boundaries
+    boundaries.zoom(width_factor, height_factor )
+  end
 
   # return [Integer, nil]
   #   sample more with small images, less with large images
   #   we want to return larger numbers (= faster sampling)
-  #
-  #
   def self.get_subdivision_size(image_width)
     case image_width
     when nil
@@ -69,6 +114,7 @@ class Sqed::BoundaryFinder
   #   (:rows|:columns), :rows finds vertical borders, :columns finds horizontal borders
   #
   def self.color_boundary_finder(image: image, sample_subdivision_size: nil, sample_cutoff_factor: nil, scan: :rows, boundary_color: :green)
+
     image_width = image.send(scan)
     sample_subdivision_size = get_subdivision_size(image_width) if sample_subdivision_size.nil?
     samples_to_take = (image_width / sample_subdivision_size).to_i - 1
@@ -105,6 +151,8 @@ class Sqed::BoundaryFinder
 
     if sample_cutoff_factor.nil?
       cutoff = max_difference(border_hits.values)
+
+      cutoff = border_hits.values.first - 1 if cutoff == 0 # difference of two identical things is 0
     else
       cutoff = (samples_to_take * sample_cutoff_factor).to_i
     end
@@ -132,6 +180,7 @@ class Sqed::BoundaryFinder
   # return [Array]
   #   the median position of all (pixel) positions that have a count greater than the cutoff
   def self.frequency_stats(frequency_hash, sample_cutoff = 0)
+   
     return nil if sample_cutoff.nil? ||  sample_cutoff < 1 
     hit_ranges = [] 
 
@@ -141,7 +190,18 @@ class Sqed::BoundaryFinder
       end
     end
 
-    return nil if hit_ranges.size < 3
+    case hit_ranges.size
+    when 1
+      c = hit_ranges[0]
+      hit_ranges = [c - 1, c, c + 1]
+    when 2
+      hit_ranges.sort!
+      c1 = hit_ranges[0]
+      c2 = hit_ranges[1]
+      hit_ranges = [c1,  c2, c2 + (c2 - c1)]
+    when 0 
+      return nil 
+    end
 
     # we have to sort because the keys (positions) we examined came unordered from a hash originally
     hit_ranges.sort!
@@ -174,3 +234,4 @@ class Sqed::BoundaryFinder
   end
 
 end
+

data/lib/sqed/boundary_finder/color_line_finder.rb

@@ -4,10 +4,15 @@ require 'rmagick'
 #
 class Sqed::BoundaryFinder::ColorLineFinder < Sqed::BoundaryFinder
 
-  def initialize(image: image, layout: layout, boundary_color: :green)
-    super(image: image, layout: layout)
+  def initialize(image: image, layout: layout, boundary_color: :green, use_thumbnail: true)
+    super(image: image, layout: layout, use_thumbnail: use_thumbnail)
     raise 'No layout provided.' if @layout.nil?
     @boundary_color = boundary_color
+
+   if use_thumbnail
+     @original_image = @img.copy
+     @img = thumbnail
+   end  
     find_bands
   end
 
@@ -25,26 +30,28 @@ class Sqed::BoundaryFinder::ColorLineFinder < Sqed::BoundaryFinder
     when :horizontal_split
       t = Sqed::BoundaryFinder.color_boundary_finder(image: img, scan: :columns, boundary_color: @boundary_color)  # set to detect horizontal division, (green line)
       return if t.nil?
+
       boundaries.set(0, [0, 0, img.columns, t[0]])  # upper section of image
       boundaries.set(1, [0, t[2], img.columns, img.rows - t[2]])  # lower section of image
 
     when :right_t   # only 3 zones expected, with horizontal division in right-side of vertical division
-      vertical = self.class.new(image: @img, layout: :vertical_split, boundary_color: @boundary_color ).boundaries
-      irt = img.crop(*vertical.for(1), true)
-      right = self.class.new(image: irt, layout: :horizontal_split, boundary_color: @boundary_color ).boundaries
+      vertical = self.class.new(image: @img, layout: :vertical_split, boundary_color: @boundary_color, use_thumbnail: false ).boundaries
 
-      boundaries.set(0, vertical.for(0))     
+      irt = img.crop(*vertical.for(1), true)
+      right = self.class.new(image: irt, layout: :horizontal_split, boundary_color: @boundary_color, use_thumbnail: false ).boundaries
+      
+      boundaries.set(0, vertical.for(0))    
       boundaries.set(1, [ vertical.x_for(1), 0, right.width_for(0), right.height_for(0) ] ) 
       boundaries.set(2, [ vertical.x_for(1), right.y_for(1), right.width_for(1), right.height_for(1)] )  
 
     when :vertical_offset_cross   # 4 zones expected, with (varying) horizontal division in left- and right- sides of vertical division
-      vertical = self.class.new(image: @img, layout: :vertical_split, boundary_color: @boundary_color ).boundaries
-    
+      vertical = self.class.new(image: @img, layout: :vertical_split, boundary_color: @boundary_color, use_thumbnail: false).boundaries
+
       ilt = img.crop(*vertical.for(0), true) 
       irt = img.crop(*vertical.for(1), true)
 
-      left = self.class.new(image: ilt, layout: :horizontal_split, boundary_color: @boundary_color ).boundaries
-      right = self.class.new(image: irt, layout: :horizontal_split, boundary_color: @boundary_color ).boundaries
+      left = self.class.new(image: ilt, layout: :horizontal_split, boundary_color: @boundary_color, use_thumbnail: false).boundaries  # fails
+      right = self.class.new(image: irt, layout: :horizontal_split, boundary_color: @boundary_color, use_thumbnail: false ).boundaries # OK
 
       boundaries.set(0, [0, 0, left.width_for(0), left.height_for(0) ]) 
       boundaries.set(1, [vertical.x_for(1), 0, right.width_for(0), right.height_for(0) ]) 
@@ -53,13 +60,13 @@ class Sqed::BoundaryFinder::ColorLineFinder < Sqed::BoundaryFinder
 
     # No specs for this yet
     when :horizontal_offset_cross
-      horizontal = self.class.new(image: @img, layout: :horizontal_split, boundary_color: @boundary_color ).boundaries
+      horizontal = self.class.new(image: @img, layout: :horizontal_split, boundary_color: @boundary_color, use_thumbnail: false ).boundaries
 
       itop = img.crop(*horizontal.for(0), true) 
       ibottom = img.crop(*horizontal.for(1), true)
 
-      top = self.class.new(image: ilt, layout: :vertical_split, boundary_color: @boundary_color ).boundaries
-      bottom = self.class.new(image: irt, layout: :vertical_split, boundary_color: @boundary_color ).boundaries
+      top = self.class.new(image: ilt, layout: :vertical_split, boundary_color: @boundary_color, use_thumbnail: false ).boundaries
+      bottom = self.class.new(image: irt, layout: :vertical_split, boundary_color: @boundary_color, use_thumbnail: false ).boundaries
 
       boundaries.set(0, [0, 0, top.width_for(0), top.height_for(0) ]) 
       boundaries.set(1, [top.x_for(1), 0, top.width_for(1), top.height_for(1) ]) 
@@ -67,8 +74,8 @@ class Sqed::BoundaryFinder::ColorLineFinder < Sqed::BoundaryFinder
       boundaries.set(3, [0, horizontal.y_for(1), bottom.width_for(0), bottom.height_for(0) ]) 
 
     when :cross # 4 zones, with perfectly intersected horizontal and vertical division
-      v = self.class.new(image: @img, layout: :vertical_split, boundary_color: @boundary_color ).boundaries
-      h = self.class.new(image: @img, layout: :horizontal_split, boundary_color: @boundary_color ).boundaries 
+      v = self.class.new(image: @img, layout: :vertical_split, boundary_color: @boundary_color, use_thumbnail: false ).boundaries
+      h = self.class.new(image: @img, layout: :horizontal_split, boundary_color: @boundary_color, use_thumbnail: false).boundaries 
   
       return if v.nil? || h.nil?
 
@@ -84,6 +91,12 @@ class Sqed::BoundaryFinder::ColorLineFinder < Sqed::BoundaryFinder
 
     boundaries.complete = true if boundaries.populated?
 
+   if use_thumbnail
+     @img = @original_image
+     zoom_boundaries
+     @original_image = nil
+   end
+
   end
 
 

data/lib/sqed/boundary_finder/cross_finder.rb

@@ -5,13 +5,13 @@ require 'rmagick'
 class Sqed::BoundaryFinder::CrossFinder < Sqed::BoundaryFinder
 
   def initialize(image: image)
-    @image = image
+    @img = image
     find_edges 
   end
 
   def find_edges
-    width = @image.columns / 2
-    height = @image.rows / 2
+    width = @img.columns / 2
+    height = @img.rows / 2
 
     boundaries.coordinates[0] = [0, 0, width, height] 
     boundaries.coordinates[1] = [width, 0, width, height] 

data/lib/sqed/boundary_finder/stage_finder.rb

@@ -3,7 +3,7 @@ require 'rmagick'
 # Some of this code was originally inspired by Emmanuel Oga's gist https://gist.github.com/EmmanuelOga/2476153.
 #
 class Sqed::BoundaryFinder::StageFinder < Sqed::BoundaryFinder
-  
+
   # The proc containing the border finding algorithim
   attr_reader :is_border
 
@@ -11,23 +11,24 @@ class Sqed::BoundaryFinder::StageFinder < Sqed::BoundaryFinder
 
   # How small we accept a cropped picture to be. E.G. if it was 100x100 and
   # ratio 0.1, min output should be 10x10
-  MIN_CROP_RATIO = 0.1    
+  MIN_CROP_RATIO = 0.1
 
-  attr_reader :x0, :y0, :x1, :y1, :min_width, :min_height, :rows, :columns 
+  attr_reader :x0, :y0, :x1, :y1, :min_width, :min_height, :rows, :columns
 
   def initialize(image: image, is_border_proc: nil, min_ratio: MIN_CROP_RATIO)
-    super(image: image, layout: :internal_box)
+    super(image: image, layout: :internal_box) 
 
-    @min_ratio =  min_ratio
+    @min_ratio = min_ratio
 
     # Initial co-ordinates
     @x0, @y0 = 0, 0
-    @x1, @y1 = img.columns, img.rows 
+    @x1, @y1 = img.columns, img.rows
     @min_width, @min_height = img.columns * @min_ratio, img.rows * @min_ratio # minimum resultant area
     @columns, @rows = img.columns, img.rows
 
+    
     # We need a border finder proc. Provide one if none was given.
-    @is_border = is_border_proc || self.class.default_border_finder(img)  # if no proc specified, use default below
+    @is_border = is_border_proc || self.class.default_border_finder(img) # if no proc specified, use default below
 
     @x00 = @x0
     @y00 = @y0
@@ -51,8 +52,9 @@ class Sqed::BoundaryFinder::StageFinder < Sqed::BoundaryFinder
   # works for 0.5, >0.137; 0.60, >0.14 0.65, >0.146; 0.70, >0.1875; 0.75, >0.1875; 0.8, >0.237; 0.85, >0.24; 0.90, >0.28; 0.95, >0.25
   # fails for 0.75, (0.18, 0.17,0.16,0.15); 0.70, 0.18;
   #
-  def self.default_border_finder(img, samples = 5, threshold = 0.75, fuzz_factor = 0.40)   # working on non-synthetic images 04-dec-2014
-   fuzz = ((Magick::QuantumRange + 1)  * fuzz_factor).to_i  
+  # this sets variables (locally) for find_edges
+  def self.default_border_finder(img, samples = 5, threshold = 0.75, fuzz_factor = 0.40) # working on non-synthetic images 04-dec-2014
+    fuzz = ((Magick::QuantumRange + 1) * fuzz_factor).to_i
     # Returns true if the edge is a border (border meaning outer region to be cropped)
     lambda do |edge|
       border, non_border = 0.0, 0.0 # maybe should be called outer, inner
@@ -79,7 +81,7 @@ class Sqed::BoundaryFinder::StageFinder < Sqed::BoundaryFinder
     # handle this exception
     return unless is_border # return if no process defined or set for @is_border
 
-    u = x1 - 1    # rightmost pixel (kind of)
+    u = x1 - 1 # rightmost pixel (kind of)
     # increment from left to right
     x0.upto(u) do |x|
       if width_croppable? && is_border[vline(x)] then
@@ -89,7 +91,7 @@ class Sqed::BoundaryFinder::StageFinder < Sqed::BoundaryFinder
       end
     end
     # increment from left to right
-    (u).downto(x0) { |x| width_croppable?  && is_border[vline(x)] ? @x1 = x - 1 : break }
+    (u).downto(x0) { |x| width_croppable? && is_border[vline(x)] ? @x1 = x - 1 : break }
 
     u = y1 - 1
     0.upto(u) do |y|
@@ -104,10 +106,11 @@ class Sqed::BoundaryFinder::StageFinder < Sqed::BoundaryFinder
 
     delta_x = 0 #width/50    # 2% of cropped image to make up for trapezoidal distortion
     delta_y = 0 #height/50   # 2% of cropped image to make up for trapezoidal distortion <- NOT 3%
-  
+
     # TODO: add conditions
-    boundaries.complete = true 
-    boundaries.coordinates[0] = [x0 + delta_x, y0 + delta_y, width - 2*delta_x, height - 2*delta_y]
+    boundaries.complete = true
+    boundaries.set(0, [x0 + delta_x, y0 + delta_y, width - 2*delta_x, height - 2*delta_y])
+
   end
 
   def width_croppable?
@@ -127,13 +130,13 @@ class Sqed::BoundaryFinder::StageFinder < Sqed::BoundaryFinder
   end
 
   # actually  + 1 (starting at zero?)
-  def width   
-    @x1 - @x0  
+  def width
+    @x1 - @x0
   end
- 
+
   # actually  + 1 (starting at zero?)
   def height
-    @y1 - @y0 
+    @y1 - @y0
   end
 
 end

data/lib/sqed/extractor.rb

@@ -42,10 +42,11 @@ class Sqed::Extractor
       if parsers = SqedConfig::SECTION_PARSERS[section_type]
 
         section_image = r.send("#{section_type}_image")
+
         updated = r.send(section_type)
 
         parsers.each do |p|
-          parsed_result = p.new(section_image).text
+          parsed_result = p.new(section_image).text(section_type: section_type)
           updated.merge!(p::TYPE => parsed_result) if parsed_result
         end
 

data/lib/sqed/parser.rb

@@ -10,8 +10,9 @@ class Sqed::Parser
     raise 'no image provided to parser' if @image && !(@image.class.name == 'Magick::Image')
   end
 
+  # TODO: is this required?!j
   # must be provided in subclasses
-  def text
+  def text(section_type: :default)
     nil
   end
 

data/lib/sqed/parser/barcode_parser.rb

@@ -36,7 +36,7 @@ class Sqed::Parser::BarcodeParser < Sqed::Parser
  #end
 
   # alias to a universal method
-  def text 
+  def text(section_type: :default) 
     barcode
   end
 

data/lib/sqed/parser/ocr_parser.rb

@@ -2,7 +2,20 @@
 #
 # Given a single image return all text in that image.
 #
-# For past reference http://misteroleg.wordpress.com/2012/12/19/ocr-using-tesseract-and-imagemagick-as-pre-processing-task/
+# For reference 
+#   http://misteroleg.wordpress.com/2012/12/19/ocr-using-tesseract-and-imagemagick-as-pre-processing-task/
+#   https://code.google.com/p/tesseract-ocr/wiki/FAQ
+#   http://www.sk-spell.sk.cx/tesseract-ocr-parameters-in-302-version
+#
+# "There is a minimum text size for reasonable accuracy. 
+# You have to consider resolution as well as point size. 
+# Accuracy drops off below 10pt x 300dpi, rapidly below 8pt x 300dpi. 
+# A quick check is to count the pixels of the x-height of your characters. 
+# (X-height is the height of the lower case x.) 
+# At 10pt x 300dpi x-heights are typically about 20 pixels, although this
+# can vary dramatically from font to font. 
+# Below an x-height of 10 pixels, you have very little chance of accurate results, 
+# and below about 8 pixels, most of the text will be "noise removed". 
 #
 require 'rtesseract' 
 
@@ -10,60 +23,84 @@ class Sqed::Parser::OcrParser < Sqed::Parser
 
   TYPE = :text
 
+  # Tesseract parameters default/specific to section type, 
+  # default is merged into the type
+  SECTION_PARAMS = {
+    default: {
+      psm: 3,
+#      classify_debug_level: 5,
+#      lang: 'eng', 
+#      load_system_dawg: 0,
+#      load_unambig_dawg: 0,
+#      load_freq_dawg: 0,
+#      load_fixed_length_dawgs: 0,
+#      load_number_dawg: 0,
+#      load_punc_dawg: 1, ## important
+#      load_unambig_dawg: 1,
+#      chop_enable: 0,
+#     enable_new_segsearch: 1,
+#     tessedit_debug_quality_metrics: 1,
+#     tessedit_write_params_to_file: 'tmp/ocr_config_file.txt',
+#     tessedit_write_images: 1,
+#     equationdetect_save_merged_image: 1,
+#     tessedit_dump_pageseg_images: 1,
+#     equationdetect_save_bi_image: 1
+    },
+    annotated_specimen: {
+      edges_children_count_limit: 3000 # was 45, significantly improves annotated_specimen for odontates
+    },
+    identifier: {
+      psm: 1,
+      # tessedit_char_whitelist: '0123456789'
+      #  edges_children_count_limit: 4000
+    }, 
+    curator_metadata: {
+    },
+    labels: {
+      psm: 3, # may need to be 6
+    },
+    deterimination_labels: {
+      psm: 3
+    }
+  }
+
   # the text extracted from the image
   attr_accessor :text
 
-  # https://code.google.com/p/tesseract-ocr/wiki/FAQ
-  def text
-    img = @image #.white_threshold(245)
-
-    # @jrflood: this is where you will have to do some research, tuning images so that they can be better ocr-ed,
-    # get potential border pixel color (based on quadrant?)
-    new_color = img.pixel_color(1, 1)
-    # img = img.scale(2)
-    # img.write('foo0.jpg.jpg')
-    # img = img.enhance
-    # img.write('foo1.jpg')
-    # img = img.quantize(8, Magick::GRAYColorspace)
-    # img.write('foo1.jpg')
-    # img = img.sharpen(1.0, 0.2)
-    # img.write('foo2.jpg')
-    # border_color = img.pixel_color(img.columns - 1, img.rows - 1)
-    # img = img.color_floodfill(img.columns - 1, img.rows - 1, new_color)
-    # img.write('tmp/foo4.jpg')
-    # img = img.quantize(2, Magick::GRAYColorspace)
-    # #img = img.threshold(0.5)
-    # img.write('foo4.jpg') # for debugging purposes, this is the image that is sent to OCR
-    # img = img.equalize #(32, Magick::GRAYColorspace)
-    # img.write('foo5.jpg') # for debugging purposes, this is the image that is sent to OCR
-    # #img.write('foo3.jpg') # for debugging purposes, this is the image that is sent to OCR
-    #
-    # img.write('foo.jpg') # for debugging purposes, this is the image that is sent to OCR
-
-
-    # From https://code.google.com/p/tesseract-ocr/wiki/FAQ
-    # " There is a minimum text size for reasonable accuracy. You have to consider resolution as well as point size. Accuracy drops off below 10pt x 300dpi, rapidly below 8pt x 300dpi. A quick check is to count the pixels of the x-height of your characters. (X-height is the height of the lower case x.) At 10pt x 300dpi x-heights are typically about 20 pixels, although this can vary dramatically from font to font. Below an x-height of 10 pixels, you have very little chance of accurate results, and below about 8 pixels, most of the text will be "noise removed". 
-
-
-    # http://www.sk-spell.sk.cx/tesseract-ocr-parameters-in-302-version
-    # doesn't supprot outputbase
-    r = RTesseract.new(img, lang: 'eng', psm: 1, 
-                       load_system_dawg: 0,
-                       tessedit_debug_quality_metrics: 1,
-                       load_freq_dawg: 1 ,
-                       chop_enable: 1,
-                       tessedit_write_images: 1,
-                       equationdetect_save_merged_image: 1,
-                       tessedit_dump_pageseg_images: 1,
-                       equationdetect_save_bi_image: 1,
-                       load_unambig_dawg: 0,
-                       tessedit_write_params_to_file: 'tmp/ocr_config_file.txt' ) # psm: 3,
-
-    # img = img.white_threshold(245)
+  # future consideration 
+  # def enhance_image(img)
+  # get potential border pixel color (based on quadrant?)
+  # new_color = img.pixel_color(1, 1)
 
+  # img = img.scale(2)
+  # img.write('foo0.jpg.jpg')
+  # img = img.enhance
+  # img.write('foo1.jpg')
+  # img = img.quantize(8, Magick::GRAYColorspace)
+  # img.write('foo1.jpg')
+  # img = img.sharpen(1.0, 0.2)
+  # img.write('foo2.jpg')
+  # border_color = img.pixel_color(img.columns - 1, img.rows - 1)
+  # img = img.color_floodfill(img.columns - 1, img.rows - 1, new_color)
+  # img.write('tmp/foo4.jpg')
+  # img = img.quantize(2, Magick::GRAYColorspace)
+  # #img = img.threshold(0.5)
+  # img.write('foo4.jpg') # for debugging purposes, this is the image that is sent to OCR
+  # img = img.equalize #(32, Magick::GRAYColorspace)
+  # img.write('foo5.jpg') # for debugging purposes, this is the image that is sent to OCR
+  # #img.write('foo3.jpg') # for debugging purposes, this is the image that is sent to OCR
+  #
+  # img.write('foo.jpg') # for debugging purposes, this is the image that is sent to OCR
+  # img = img.white_threshold(245)
+  # img
+  # end
+  
+  def text(section_type: :default)
+    img = @image 
+    params = SECTION_PARAMS[:default].merge(SECTION_PARAMS[section_type])
+    r = RTesseract.new(img, params) 
     @text = r.to_s.strip 
   end
 
-  # Need to provide tuning methods here, i.e. image transormations that facilitate OCR
 
 end

data/lib/sqed/version.rb

@@ -1,3 +1,3 @@
 class Sqed
-  VERSION = "0.1.5"
+  VERSION = "0.1.6"
 end

data/spec/lib/sqed/boundary_finder/color_line_finder_spec.rb

@@ -8,25 +8,25 @@ describe Sqed::BoundaryFinder::ColorLineFinder do
   let(:c) {b.boundaries}
   let(:d) { image.crop(*c.for(0), true) }
   
-  let(:e) { Sqed::BoundaryFinder::ColorLineFinder.new(image: d, layout: :right_t) }
+  let(:e) { Sqed::BoundaryFinder::ColorLineFinder.new(image: d, layout: :right_t, use_thumbnail: false) }
   let(:f) { e.boundaries }
-  let(:g) { Sqed::BoundaryFinder::ColorLineFinder.new(image: d, layout: :vertical_offset_cross)}
+  let(:g) { Sqed::BoundaryFinder::ColorLineFinder.new(image: d, layout: :vertical_offset_cross, use_thumbnail: false)}
   let(:h) { g.boundaries }
-  let(:gv) { Sqed::BoundaryFinder::ColorLineFinder.new(image: d, layout: :vertical_split) }
+  let(:gv) { Sqed::BoundaryFinder::ColorLineFinder.new(image: d, layout: :vertical_split, use_thumbnail: false) }
   let(:hv) { gv.boundaries }
  
   let(:ah) { ImageHelpers.vertical_offset_cross_red }
   let(:bh) { Sqed::BoundaryFinder::StageFinder.new(image: ah) }
   let(:ch) { bh.boundaries }
   let(:dh) { ah.crop(*ch.for(0), true) }
-  let(:gh) { Sqed::BoundaryFinder::ColorLineFinder.new(image: dh, layout: :horizontal_split, boundary_color: :red) } # was :horizontal_split
+  let(:gh) { Sqed::BoundaryFinder::ColorLineFinder.new(image: dh, layout: :horizontal_split, boundary_color: :red, use_thumbnail: false) } # was :horizontal_split
   let(:hh) { gh.boundaries }
 
   let(:ibs) { ImageHelpers.black_stage_green_line_specimen } 
   let(:bbs) { Sqed::BoundaryFinder::StageFinder.new(image: ibs) } 
   let(:cbs) { bbs.boundaries }
   let(:dbs) { ibs.crop(*cbs.for(0), true) }
-  let(:gbs) { Sqed::BoundaryFinder::ColorLineFinder.new(image: dbs, layout: :vertical_offset_cross) }
+  let(:gbs) { Sqed::BoundaryFinder::ColorLineFinder.new(image: dbs, layout: :vertical_offset_cross, use_thumbnail: false) }
   let(:hbs) { gbs.boundaries }
 
   specify 'initial image columns are as expected for :image above' do
@@ -167,7 +167,7 @@ describe Sqed::BoundaryFinder::ColorLineFinder do
   context 'thumbnail processing finds reasonable boundaries' do
 
     let(:thumb) { ImageHelpers.frost_stage_thumb  }
-    let(:finder) { Sqed::BoundaryFinder::ColorLineFinder.new(image: thumb, layout: :cross)}
+    let(:finder) { Sqed::BoundaryFinder::ColorLineFinder.new(image: thumb, layout: :cross, use_thumbnail: false)}
     let(:finder_boundaries) { finder.boundaries }
 
     let(:pct)  { 0.08 }

data/spec/lib/sqed/boundary_finder/cross_finder_spec.rb

@@ -22,7 +22,8 @@ describe Sqed::BoundaryFinder::CrossFinder do
   end
 
   specify 'the 0th image has height = 300' do
-    expect(c.height_for(0)).to eq(300)
+    pct = 0.02
+    expect(c.height_for(0)).to be_within(pct*300).of(300)
   end
 
 end 

data/spec/lib/sqed/boundary_finder_spec.rb

@@ -3,7 +3,7 @@ require 'spec_helper'
 describe Sqed::BoundaryFinder do
 
   specify 'when no image provided, #new raises' do
-    expect { Sqed::BoundaryFinder.new() }.to raise_error
+    expect { Sqed::BoundaryFinder.new() }.to raise_error('No layout provided.')
   end
 
   context 'when initiated with an image' do
@@ -103,9 +103,15 @@ describe Sqed::BoundaryFinder do
       expect( Sqed::BoundaryFinder.frequency_stats(i, 12)).to eq([3, 4, 5])
     end
 
+    specify 'returns estimated borders if only one hit greater than samples taken' do
+      expect( Sqed::BoundaryFinder.frequency_stats(i, 15)).to eq([2,3,4])
+    end
+
     specify 'returns nil if no count is greater than samples taken' do
-      expect( Sqed::BoundaryFinder.frequency_stats(i, 15)).to eq(nil)
+      expect( Sqed::BoundaryFinder.frequency_stats(i, 20)).to eq(nil)
     end
+ 
+
   end
 
   context 'offset boundaries from crossy_black_line_specimen image ' do

data/spec/lib/sqed_spec.rb

@@ -90,20 +90,23 @@ describe Sqed do
 
     context '#result' do
       let(:r) { s.result }
+
       specify 'returns a Sqed::Result' do
         expect(r.class.name).to eq('Sqed::Result')
       end
 
       context 'extracted data' do
-        specify 'for an :identifier section' do
+        specify 'text for an :identifier section' do
+
+          r.identifier_image.write('41.jpg')
           expect(r.text_for(:identifier)).to match('000041196')
         end
 
-        specify 'for an annotated_specimen section' do
+        specify 'text for an annotated_specimen section' do
           expect(r.text_for(:annotated_specimen)).to match('Saucier Creek')
         end
 
-        specify 'for a curator_metadata section' do
+        specify 'text for a curator_metadata section' do
           expect(r.text_for(:curator_metadata)).to match('Frost Entomological Museum')
         end
       end
@@ -113,7 +116,7 @@ describe Sqed do
   context 'all together, with border' do
     let(:image) { ImageHelpers.greenline_image }
     let(:pattern) { :right_t }
-    let(:s) { Sqed.new(image: image, pattern: pattern, has_border: false)  }
+    let(:s) { Sqed.new(image: image, pattern: pattern, has_border: true)  }
 
     specify '#boundaries returns a Sqed::Boundaries instance' do
       expect(s.boundaries.class.name).to eq('Sqed::Boundaries')
@@ -138,11 +141,12 @@ describe Sqed do
       end
 
       context 'extracted data' do
-        specify 'for an :identifier section' do
+        specify 'text for an :identifier section' do
+          r.identifier_image.write('85.jpg')
           expect(r.text_for(:identifier)).to match('000085067')
         end
 
-        specify 'for a specimen section' do
+        specify 'text for a specimen section' do
           expect(r.text_for(:annotated_specimen)).to match('Aeshna')
         end
       end

data/sqed.gemspec

@@ -23,7 +23,7 @@ Gem::Specification.new do |spec|
   spec.add_dependency 'rtesseract', '~> 1.2.6'
   spec.add_dependency 'zxing_cpp', '~> 0.1.0'
 
-  spec.add_development_dependency 'rspec'
+  spec.add_development_dependency 'rspec', '~> 3.3'
   spec.add_development_dependency 'bundler', '~> 1.5'
   spec.add_development_dependency 'did_you_mean', '~> 0.9'
   spec.add_development_dependency 'byebug'

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: sqed
 version: !ruby/object:Gem::Version
-  version: 0.1.5
+  version: 0.1.6
 platform: ruby
 authors:
 - Matt Yoder
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-06-11 00:00:00.000000000 Z
+date: 2015-09-09 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -71,16 +71,16 @@ dependencies:
   name: rspec
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '0'
+        version: '3.3'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '0'
+        version: '3.3'
 - !ruby/object:Gem::Dependency
   name: bundler
   requirement: !ruby/object:Gem::Requirement