sqed 0.0.4 → 0.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: bed58b448dbb4cd6681353d53ecf41a5d5bc86d3
-  data.tar.gz: 2a59cd7ffd1033a230d37eca146f40c9ced0bb77
+  metadata.gz: 4600b58a8dca4c59c21e0fcaa08c63e7564a56f4
+  data.tar.gz: f1f96f8e93e988bd53ca354c4fe3a67f359b1368
 SHA512:
-  metadata.gz: 0bf133d7ff4b70b1f064a3450b36f9db622c84d0a104e1ae965e40c23c5d8e8b1711c6724dc7af297a5bf864fecaa2d27a9c6f2ec8e40c812bdf5049b5ac8c45
-  data.tar.gz: c7b7ad499e5a39daa8f71144fa363a8af69c5dec95842ee7de35bcf4d6e72fbebca88f4f84313cfb7c5f87e9467fe828ce79539e175057c07f4e77d6cd0875fb
+  metadata.gz: d7cb582b1d6db468617a132ce67a0dccab2315911eb650468ef6d0e756333cfcca710969d9d845dbe8c3149059499c486845072a053f5674412af268928e2f72
+  data.tar.gz: 9a24446ca6fb5caf9d7a9c436013e31cc6ceeb09e4c06f1968ee4cb5a2d544bd939d7b59e3bda023b2cfea38c4eb0cb1182e0971290738c44c8135a33d046b08

data/.gitignore

@@ -24,4 +24,6 @@ local/
 foo*.jpg
 SessionID_BarcodeImage.JPG
 
+tess*
+
 /*.jpg

data/README.md

@@ -3,7 +3,7 @@
 
 # Sqed
 
-Stub for a gem that supports specimen digitization from images.
+Sqed is a gem that faciliates metadata extraction from images of staged collection objects. 
 
 ## Installation
 
@@ -21,7 +21,7 @@ Or install it yourself as:
 
 ## Usage
 
-TODO: Write usage instructions here
+For the time being, see specs.
 
 ## Contributing
 

data/lib/sqed/boundaries.rb

@@ -12,13 +12,15 @@ class Sqed::Boundaries
   #   0 => [10,10,40,40]
   attr_reader :coordinates
 
-  # An Sqed::Config::EXTRACTION_PATTERN layout
+  # A symbol from Sqed::Config::LAYOUTS.keys 
+  #   :right_t
   attr_accessor :layout
 
-  # Whether or not the last method to populate this object passed fully
+  # Boolean, whether or not the last method to populate this object passed fully
   attr_accessor :complete
 
   def initialize(layout = nil)
+    raise 'unrecognized layout' if layout && !SqedConfig::LAYOUTS.include?(layout)
     @complete = false
 
     @layout = layout
@@ -35,11 +37,12 @@ class Sqed::Boundaries
   def offset(boundary)
     b = Sqed::Boundaries.new()    # the idea here is to create a deep copy of self, offsetting by boundary as we go
     (0..self.coordinates.length - 1).each do |i|
-      b.coordinates[i] = []  # create the instance of the i-th coordinate, then populate it
-      b.coordinates[i][0] = self.x_for(i) + boundary.x_for(0)
-      b.coordinates[i][1] = self.y_for(i) + boundary.y_for(0)
-      b.coordinates[i][2] = self.width_for(i)
-      b.coordinates[i][3] = self.height_for(i)
+      b.set(i, 
+            [(self.x_for(i) + boundary.x_for(0)),
+             (self.y_for(i) + boundary.y_for(0)),
+             self.width_for(i),
+             self.height_for(i)]
+           )
     end
     b.complete = self.complete
     b
@@ -76,4 +79,16 @@ class Sqed::Boundaries
     @coordinates[index][3]
   end
 
+  def set(index, coordinates)
+    @coordinates[index] = coordinates
+  end
+
+  def populated?
+      @coordinates.each do |c|
+        return false if c[0].nil?
+      end
+    true 
+  end
+
+
 end

data/lib/sqed/boundary_finder/color_line_finder.rb

@@ -19,15 +19,15 @@ class Sqed::BoundaryFinder::ColorLineFinder < Sqed::BoundaryFinder
     when :vertical_split    # can vertical and horizontal split be re-used to do cross cases?
       t = Sqed::BoundaryFinder.color_boundary_finder(image: img, boundary_color: @boundary_color)  #detect vertical division, green line
       return if t.nil?
-      boundaries.coordinates[0] = [0, 0, t[0], img.rows]  # left section of image
-      boundaries.coordinates[1] = [t[2], 0, img.columns - t[2], img.rows]  # right section of image
+      boundaries.set(0, [0, 0, t[0], img.rows])  # left section of image
+      boundaries.set(1, [t[2], 0, img.columns - t[2], img.rows])  # right section of image
       boundaries.complete = true
 
     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.coordinates[0] = [0, 0, img.columns, t[0]]  # upper section of image
-      boundaries.coordinates[1] = [0, t[2], img.columns, img.rows - t[2]]  # lower section of image
+      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
       boundaries.complete = true
       # boundaries.coordinates[2] = [0, 0, img.columns, t[1]]  # upper section of image
       # boundaries.coordinates[3] = [0, t[1], img.columns, img.rows - t[1]]  # lower section of image
@@ -35,45 +35,52 @@ class Sqed::BoundaryFinder::ColorLineFinder < Sqed::BoundaryFinder
     when :right_t   # only 3 zones expected, with horizontal division in right-side of vertical division
       t = Sqed::BoundaryFinder.color_boundary_finder(image: img, boundary_color: @boundary_color)  #defaults to detect vertical division, green line
       return if t.nil?
-      boundaries.coordinates[0] = [0, 0, t[0], img.rows]  # left section of image
-      boundaries.coordinates[1] = [t[2], 0, img.columns - t[2], img.rows]  # left section of image
+
+      left = [0, 0, t[0], img.rows]
+      right = [t[2], 0, img.columns - t[2], img.rows]
+
+      boundaries.set(0, left)            # left section of image
 
       # now subdivide right side
-      irt = img.crop(*boundaries.coordinates[1], true)
+      irt = img.crop(*right, true)
       rt = Sqed::BoundaryFinder.color_boundary_finder(image: irt, scan: :columns, boundary_color: @boundary_color)  # set to detect horizontal division, (green line)
       return if rt.nil?
-      boundaries.coordinates[1] = [t[2], 0, img.columns - t[2], rt[0]]  # upper section of image
-      boundaries.coordinates[2] = [t[2], rt[2], img.columns - t[2], img.rows - rt[2]]  # lower section of image
+      boundaries.set(1, [t[2], 0, img.columns - t[2], rt[0]])                # upper section of image
+      boundaries.set(2, [t[2], rt[2], img.columns - t[2], img.rows - rt[2]]) # lower section of image
       boundaries.complete = true
       # will return 1, 2, or 3
 
     when :offset_cross   # 4 zones expected, with horizontal division in right- and left- sides of vertical division
       t = Sqed::BoundaryFinder.color_boundary_finder(image: img, boundary_color: @boundary_color)  # defaults to detect vertical division, green line
       raise if t.nil?
-      boundaries.coordinates[0] = [0, 0, t[0], img.rows]  # left section of image
-      boundaries.coordinates[1] = [t[2], 0, img.columns - t[2], img.rows]  # right section of image
+
+      left = [0, 0, t[0], img.rows]                               # left section of image
+      right = [t[2], 0, img.columns - t[2], img.rows]             # right section of image
 
       # now subdivide left side
-      ilt = img.crop(*boundaries.coordinates[0], true)
+      ilt = img.crop(*left, true)
 
       lt = Sqed::BoundaryFinder.color_boundary_finder(image: ilt, scan: :columns, boundary_color: @boundary_color)  # set to detect horizontal division, (green line)
+
       if !lt.nil?
-        boundaries.coordinates[0] = [0, 0, t[0], lt[0]]  # upper section of image
-        boundaries.coordinates[3] = [0, lt[2], t[0], img.rows - lt[2]]  # lower section of image
+        boundaries.set(0, [0, 0, left[2], lt[0]])                 # upper section of image
+        boundaries.set(3, [0, lt[2], left[2], img.rows - lt[2]])  # lower section of image
       end
 
       # now subdivide right side
-      irt = img.crop(*boundaries.coordinates[1], true)
+      irt = img.crop(*right, true)
       rt = Sqed::BoundaryFinder.color_boundary_finder(image: irt, scan: :columns, boundary_color: @boundary_color)  # set to detect horizontal division, (green line)
       return if rt.nil?
 
-      boundaries.coordinates[1] = [t[2], 0, img.columns - t[2], rt[0]]  # upper section of image
-      boundaries.coordinates[2] = [t[2], rt[2], img.columns - t[2], img.rows - rt[2]]  # lower section of image
+      boundaries.set(1, [t[2], 0, img.columns - t[2], rt[0]])                 # upper section of image
+      boundaries.set(2, [t[2], rt[2], img.columns - t[2], img.rows - rt[2]])  # lower section of image
       # will return 1, 2, 3, or 4  //// does not handle staggered vertical boundary case
-      boundaries.complete = true
+      
+      boundaries.complete = true if boundaries.populated?
 
     else
-      boundaries.coordinates[0] = [0, 0, img.columns, img.rows]  # totality of image as default
+      boundaries.set(0, [0, 0, img.columns, img.rows])   # totality of image as default
+      # TODO: boundaries.complete status here?
       return    # return original image boundary if no method implemented
     end
 

data/lib/sqed/boundary_finder.rb

@@ -52,9 +52,9 @@ class Sqed::BoundaryFinder
     (0..samples_to_take).each do |s|
       # Create a sample image a single pixel tall
       if scan == :rows
-        j = image.crop(0, s * sample_subdivision_size, image.columns, 1)
+        j = image.crop(0, s * sample_subdivision_size, image.columns, 1, true)
       elsif scan == :columns
-        j = image.crop(s * sample_subdivision_size, 0, 1, image.rows)
+        j = image.crop(s * sample_subdivision_size, 0, 1, image.rows, true)
       else
         raise
       end

data/lib/sqed/extractor.rb

@@ -1,45 +1,61 @@
 require 'RMagick'
 
-# An Extractor takes Boundries object and a layout pattern and returns a Sqed::Result
+# An Extractor takes Boundries object and a metadata_map pattern and returns a Sqed::Result
 # 
 class Sqed::Extractor
 
-  attr_accessor :boundaries, :layout, :image
+  # a Sqed::Boundaries instance 
+  attr_accessor :boundaries
 
-  def initialize(boundaries: boundaries, layout: layout, image: image)
-    raise if boundaries.nil? || !boundaries.class == Sqed::Boundaries
-    raise if layout.nil? || !layout.class == Hash
+  # a metadata_map hash from EXTRACTION_PATTERNS like:
+  #   {0 => :annotated_specimen, 1 => :identifier, 2 =>:image_registration }
+  attr_accessor :metadata_map
 
-    @layout = layout
+  # a Magick::Image file
+  attr_accessor :image
+
+  def initialize(boundaries: boundaries, metadata_map: metadata_map, image: image)
+    raise 'boundaries not provided or provided boundary is not a Sqed::Boundaries' if boundaries.nil? || !boundaries.class == Sqed::Boundaries
+    raise 'metadata_map not provided or metadata_map not a Hash' if metadata_map.nil? || !metadata_map.class == Hash
+    raise 'image not provided' if image.nil? || !image.class == Magick::Image
+
+    @metadata_map = metadata_map
     @boundaries = boundaries
     @image = image
   end
 
   def result
     r = Sqed::Result.new()
-   
+
     # assign the images to the result
-    boundaries.each do |section, coords|
-      r.send("#{LAYOUT_SECTION_TYPES[section]}=", extract_image(coords))
+    boundaries.each do |section_index, coords|
+      image_setter = "#{metadata_map[section_index]}_image="
+      r.send(image_setter, extract_image(coords))
     end 
 
     # assign the metadata to the result
-    layout.keys.each do |section_index, section_type|
+    metadata_map.each do |section_index, section_type|
       # only extract data if a parser exists
-      if parser = SECTION_PARSERS[section_type]
-        r.send("#{section_type}=", parser.new(image: r.send(section_type + "_image").text) )
+      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
+          updated.merge!(p::TYPE => parsed_result) if parsed_result
+        end
+
+        r.send("#{section_type}=", updated) 
       end
     end
 
     r
   end
 
-  # coords are x1, y1, x2, y2
+  # crop takes x, y, width, height
   def extract_image(coords)
-    # crop takes x, y, width, height
-    # @image.crop(coords[0], coords[1], coords[2] - coords[0], coords[3] - coords[1] )
-    bp = 0
-    @image.crop(coords[0], coords[1], coords[2], coords[3], true)
+    i = @image.crop(*coords, true)
   end
 
 end

data/lib/sqed/parser/barcode_parser.rb

@@ -1,27 +1,43 @@
 # Given an image, return an ordered array of detectable barcodes
 
+
+
 class Sqed::Parser::BarcodeParser < Sqed::Parser
-  attr_accessor :barcodes
+
+  TYPE = :barcode
+
+  attr_accessor :image
+
+  attr_accessor :barcode
 
   def initialize(image)
     super
-    @barcodes = bar_codes
+    @image = image
   end
 
-  def bar_codes
-    # process the images, spit out the barcodes
-    # return ZXing.decode_all(@image)   #['ABC 123', 'DEF 456']
-    # a = `/usr/local/Cellar/zbar/0.10_1/bin/zbarimg ~/src/sqed/spec/support/files/test_barcode.JPG`
-    # b = a.split("\n")
-    f = 'SessionID_BarcodeImage.JPG'
-    i = @image[:image]
-    if i.nil?
-      i = @image
-    end
-    i.write("tmp/#{f}")
-    c = `/usr/local/Cellar/zbar/0.10_1/bin/zbarimg #{f}`
-    d = c.split("\n")
-    return d
+  def barcode
+    @barcode ||= get_barcode
+    @barcode
+  end
+
+  # Uses the same enging as zbarimg that you can install with brew (zbarimg)
+  #
+  def get_code_128
+    ZXing.decode @image.filename
+  end
+
+  # try a bunch of options, organized by most common,  give the first hit
+  def get_barcode
+    [get_code_128].compact.first
+  end 
+
+ #def get_datamatrix
+ #  https://github.com/srijan/ruby-dmtx
+ #end
+
+  # alias to a universal method
+  def text 
+    barcode
   end
 
 end

data/lib/sqed/parser/ocr_parser.rb

@@ -7,8 +7,13 @@
 require 'rtesseract' 
 
 class Sqed::Parser::OcrParser < Sqed::Parser
-  attr_accessor  :text
 
+  TYPE = :text
+
+  # 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)
 
@@ -19,9 +24,6 @@ class Sqed::Parser::OcrParser < Sqed::Parser
     # img = img.scale(2)
     # img.write('foo0.jpg.jpg')
     # img = img.enhance
-    # img = img.enhance
-    # img = img.enhance
-    # img = img.enhance
     # img.write('foo1.jpg')
     # img = img.quantize(8, Magick::GRAYColorspace)
     # img.write('foo1.jpg')
@@ -39,12 +41,28 @@ class Sqed::Parser::OcrParser < Sqed::Parser
     #
     # img.write('foo.jpg') # for debugging purposes, this is the image that is sent to OCR
 
-    r = RTesseract.new(img, lang: 'eng', psm: 3)
 
+    # 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)
 
-    @text = r.to_s 
+    @text = r.to_s.strip 
   end
 
   # Need to provide tuning methods here, i.e. image transormations that facilitate OCR

data/lib/sqed/parser.rb

@@ -7,5 +7,12 @@ class Sqed::Parser
 
   def initialize(image)
     @image = image 
+    raise 'no image provided to parser' if @image && !(@image.class.name == 'Magick::Image')
   end
+
+  # must be provided in subclasses
+  def text
+    nil
+  end
+
 end

data/lib/sqed/result.rb

@@ -1,15 +1,63 @@
-
-# A Sqed::Result is a wrapper for the results of the
-# full process of data extraction from an image.
-#
-#
+# A Sqed::Result is a container for the results of the
+# the data extraction for the full stage 
 #
 class Sqed::Result
 
   SqedConfig::LAYOUT_SECTION_TYPES.each do |k|
-    attr_accessor k
     attr_accessor "#{k}_image".to_sym
+    attr_accessor k
+  end
+
+  def initialize
+    SqedConfig::LAYOUT_SECTION_TYPES.each do |k|
+      send("#{k}=", {}) 
+    end
+  end
+
+  # return [String, nil]
+  #   the text derived from the OCR parsing of the section
+  def text_for(section)
+    send(section)[:text] 
   end
+
+  # return [String, nil]
+  #   the text derived from the barcode parsing of the section
+  def barcode_text_for(section)
+    send(section)[:barcode]
+  end
+
+  # return [Hash]
+  #   a map of layout_section_type => value (if there is a value), 
+  #   i.e. all possible parsed text values returned from the parser
+  def text
+    result = {} 
+    SqedConfig::LAYOUT_SECTION_TYPES.each do |k|
+      v = self.send(k)
+      result.merge!(k => v) if v[:barcode] || v[:text] 
+    end
+    result
+  end
+
+  # return [Hash]
+  #   a map of layout_section_type => Image
+  def images
+    result = {} 
+    SqedConfig::LAYOUT_SECTION_TYPES.each do |k|
+      image = self.send("#{k}_image")
+      result.merge!(k => image) if image
+    end
+    result
+  end
+
+  # return [True]
+  #   write the images in #images to tmp/ 
+  def write_images
+    images.each do |k, img|
+      img.write("tmp/#{k}.jpg")
+    end
+    true
+  end
+
 end
 
 

data/lib/sqed/version.rb

@@ -1,3 +1,3 @@
 class Sqed
-  VERSION = "0.0.4"
+  VERSION = "0.1.0"
 end

data/lib/sqed.rb

@@ -4,7 +4,7 @@ recent_ruby = RUBY_VERSION >= '2.1.1'
 raise "IMPORTANT: sqed gem requires ruby >= 2.1.1" unless recent_ruby
 
 require "RMagick"
-# include Magick
+include Magick
 
 # Instants take the following
 # 1) A base image @image
@@ -28,19 +28,27 @@ class Sqed
   attr_accessor :pattern
 
   # the image that is the cropped content for parsing
-  attr_accessor :stage_image, :stage_boundary, :boundaries, :auto_detect_border, :boundary_color
+  attr_accessor :stage_image
+
+  # a Sqed::Boundaries instance that stores the coordinates of the stage 
+  attr_accessor :stage_boundary
+ 
+  # a Sqed::Boundaries instances that contains the coordinates of the interan stage sections
+  attr_accessor :boundaries
+ 
+  # Boolean, whether to detect the border on initialization, i.e. new()
+  attr_accessor :auto_detect_border 
+ 
+  # a symbol, :red, :green, :blue, describing the boundary color within the stage 
+  attr_accessor :boundary_color
 
   def initialize(image: image, pattern: pattern, auto_detect_border: true, boundary_color: :green)
     @image = image
-
     @boundaries = nil
-    @stage_boundary = Sqed::Boundaries.new(:internal_box) # a.k.a. stage
-
+    @stage_boundary = Sqed::Boundaries.new(:internal_box) 
     @auto_detect_border = auto_detect_border
-
     @pattern = pattern
     @pattern ||= :standard_cross
-
     @boundary_color = boundary_color
 
     set_stage_boundary if @auto_detect_border && @image
@@ -60,6 +68,8 @@ class Sqed
     @boundaries
   end
 
+  # Return [Sqed::Boundaries instance]
+  #   a boundaries instance that has the original image (prior to cropping stage) coordinates
   def native_boundaries
     # check for @boundaries.complete first? OR handle partial detections  ?!
     if @boundaries.complete
@@ -69,14 +79,18 @@ class Sqed
     end 
   end
 
+  # return [Image]
+  #   crops the image if not already done
   def stage_image
     crop_image if @stage_boundary.complete && @stage_image.nil?
     @stage_image 
   end
 
+  # return [Image]
+  #   crops the stage if not done, then sets/returns @stage_image
   def crop_image
     if @stage_boundary.complete
-      @stage_image = @image.crop(*@stage_boundary.for(SqedConfig.index_for_section_type(:stage, :stage)))
+      @stage_image = @image.crop(*@stage_boundary.for(SqedConfig.index_for_section_type(:stage, :stage)), true)
     else
       @stage_image = @image 
     end
@@ -84,14 +98,24 @@ class Sqed
 
   def result
     return false if @image.nil? || @pattern.nil? 
-    crop_image
     extractor = Sqed::Extractor.new(
-      boundaries: @boundaries,
-      layout: SqedConfig::EXTRACTION_PATTERNS[@pattern][:layout],
-      image: @stage_image)
+      boundaries: boundaries,
+      metadata_map: SqedConfig::EXTRACTION_PATTERNS[@pattern][:metadata_map],
+      image: stage_image)
     extractor.result
   end
 
+  def attributes
+    {
+      image: @image,
+      boundaries: @boundaries,
+      stage_boundary: @stage_boundary,
+      auto_detect_border: @auto_detect_border,
+      pattern: @pattern,
+      boundary_color: @boundary_color
+    }
+  end
+
   protected
 
   def set_stage_boundary
@@ -103,6 +127,7 @@ class Sqed
 
   def get_section_boundaries
     boundary_finder_class = SqedConfig::EXTRACTION_PATTERNS[@pattern][:boundary_finder]
+
     options = {image: stage_image}
     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'

data/lib/sqed_config.rb

@@ -1,5 +1,7 @@
 # encoding: UTF-8
 
+require 'zxing'
+
 require_relative "sqed/parser"
 require_relative "sqed/parser/ocr_parser"
 require_relative "sqed/parser/barcode_parser"
@@ -51,7 +53,6 @@ module SqedConfig
 
   # Hash values are used to stub out
   # the Sqed::Boundaries instance.
-  #
   LAYOUTS = {
     cross: [0,1,2,3],
     offset_cross: [0,1,2,3],
@@ -62,7 +63,7 @@ module SqedConfig
     internal_box: [0]
   }
 
-  #  Each element of the layout is a "section".  
+  # Each element of the layout is a "section".  
   LAYOUT_SECTION_TYPES = [
     :stage,                 # the image contains the full stage 
     :specimen,              # the specimen only, no metadata should be present
@@ -70,26 +71,31 @@ module SqedConfig
     :determination_labels,  # the section contains text that determines the specimen
     :labels,                # the section contains collecting event and non-determination labels
     :identifier,            # the section contains an identifier (e.g. barcode or unique number)
-    :image_registration     # the section contains only image registration information
+    :image_registration,    # the section contains only image registration information,
+    :curator_metadata,      # the section contains text with curator metadata
+    :nothing                # section is empty 
   ] 
 
   # Links section types to data parsers
   SECTION_PARSERS = {
-    labels: Sqed::Parser::OcrParser,
-    identifier: Sqed::Parser::BarcodeParser,
-    deterimination_labels: Sqed::Parser::OcrParser
+    labels: [ Sqed::Parser::OcrParser ],
+    identifier: [ Sqed::Parser::BarcodeParser, Sqed::Parser::OcrParser ],
+    deterimination_labels: [ Sqed::Parser::OcrParser ],
+    curator_metadata: [  Sqed::Parser::OcrParser ],
+    specimen: [ Sqed::Parser::OcrParser ],
   }
 
   EXTRACTION_PATTERNS = {
     right_t: {
       boundary_finder: Sqed::BoundaryFinder::ColorLineFinder,
       layout: :right_t,
-      metadata_map: {0 => :annotated_specimen, 1 => :identifiers, 2 =>:image_registration }
+      metadata_map: {0 => :annotated_specimen, 1 => :identifier, 2 =>:image_registration }
     },
     offset_cross: {
       boundary_finder: Sqed::BoundaryFinder::ColorLineFinder,
       layout: :offset_cross,
-      metadata_map: {0 => :annotated_specimen, 1 => :identifiers, 2 =>:image_registration }
+      metadata_map: {0 => :curator_metadata, 1 => :identifier, 2 => :image_registration, 3 => :specimen }
+     # metadata_map: {0 => :annotated_specimen, 1 => :identifier, 2 =>:image_registration }
     },
     standard_cross: {
       boundary_finder: Sqed::BoundaryFinder::CrossFinder,