sqed 0.1.7 → 0.1.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: bc7181ce86a616f52763dc679ac56b8a61757af2
-  data.tar.gz: adad2a5e8268883aa92773ca9c9bb7e7388993b6
+  metadata.gz: 639226420485cffa59f748b6c0e5c66db21b239e
+  data.tar.gz: 425bd2c26e16339e4ef7b61b605f90bf2efb42db
 SHA512:
-  metadata.gz: 48aad3560defbe2d110b108e137150bd00642fee93c7d3fdc230617395ee6e9c401d8bd1ed03164e509ff036fffad44cff4e46867d5e8125c8ccf88da7d2ab21
-  data.tar.gz: f86ca07a14ffab425ce2dad714f3a4d8a3d0676b62ca2abcba6b7c7d199bec580b8dbcc4fe2b83894bb9031123b9c49d40eff99fe92d9076643cb7fe0be59760
+  metadata.gz: 0b40ba1db0b861fb88769d99997c840931f916dacd1731b0f321828c54fbe033e552c879298a0e2a5057b3859a534487c298e8870fbd4ea7316b3351e95a4504
+  data.tar.gz: 7bd37213c089d409778291ad30aa63c5c65a0ebdd10a311ed376388b9eddc2134da234b788600b7d4213042c6290ccb88a120969d5de08aed9571cd9d5c2a2f6

data/lib/sqed.rb

@@ -77,13 +77,13 @@ class Sqed
   #   federate extraction options and apply user provided over-rides
   def extraction_metadata
     data = SqedConfig::EXTRACTION_PATTERNS[@pattern]
-
+    
+    data.merge!(boundary_color: boundary_color) 
     data.merge!(boundary_finder: @boundary_finder) if boundary_finder
+    data.merge!(has_border: has_border) 
     data.merge!(layout: layout) if layout 
     data.merge!(metadata_map: metadata_map) if metadata_map
-    data.merge!(has_border: has_border) 
     data.merge!(use_thumbnail: use_thumbnail) 
-    data.merge!(boundary_color: boundary_color) 
     data
   end
 
@@ -133,6 +133,7 @@ class Sqed
   end
 
   def result
+    # pattern.nil? is no longer true -> must have values for all extraction_metadata keys
     return false if image.nil? || pattern.nil? 
     extractor = Sqed::Extractor.new(
       boundaries: boundaries,

data/lib/sqed/boundary_finder.rb

@@ -95,8 +95,8 @@ class Sqed::BoundaryFinder
     end
   end
 
-  # @return
-  #   the column (x position) in the middle of the single green vertical line dividing the stage
+  # @return [Array]
+  #   the x or y position returned as a start, mid, and end coordinate that represent the width  of the colored line that completely divides the image, e.g. [9, 15, 16] 
   #
   # @param image
   #   the image to sample
@@ -178,7 +178,7 @@ class Sqed::BoundaryFinder
   end
 
   # return [Array]
-  #   the median position of all (pixel) positions that have a count greater than the cutoff
+  #   the start, mid, endpoint 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 

data/lib/sqed/boundary_finder/color_line_finder.rb

@@ -1,6 +1,6 @@
 require 'rmagick'
 
-# This was "green" line finder attempting to be agnostic; now it is reworked to be color-specific line finder
+# An agnostic pattern finder for color-line delimited boundaries
 #
 class Sqed::BoundaryFinder::ColorLineFinder < Sqed::BoundaryFinder
 
@@ -21,20 +21,45 @@ class Sqed::BoundaryFinder::ColorLineFinder < Sqed::BoundaryFinder
   def find_bands
     case @layout    # boundaries.coordinates are referenced from stage image
 
-    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
+    # No specs for this yet
+    when :seven_slot
+      top_bottom_split = Sqed::BoundaryFinder.color_boundary_finder(image: img, scan: :columns, boundary_color: @boundary_color)              # detect vertical division [array]
+      left_right_split = Sqed::BoundaryFinder.color_boundary_finder(image: img, sample_subdivision_size: 2, boundary_color: @boundary_color)  # detect horizontal division [array]
+
+      boundaries.set(0, [0, 0, left_right_split[0], top_bottom_split[0] ])  
+      boundaries.set(6, [0, top_bottom_split[2], left_right_split[0], img.rows - top_bottom_split[2] ] )  
+
+      right_top_img = img.crop( left_right_split[2], 0, img.columns - left_right_split[2], top_bottom_split[0] , true) # sections 1,2
+      right_bottom_img = img.crop(left_right_split[2], top_bottom_split[2], img.columns - left_right_split[2], img.rows - top_bottom_split[2], true)  # sections 3,4,5
+
+      right_top_split = Sqed::BoundaryFinder.color_boundary_finder(image: right_top_img, boundary_color: @boundary_color) # vertical line b/w 1 & 2                   
+      
+      boundaries.set(1, [left_right_split[2], 0, left_right_split[2] + right_top_split[0], top_bottom_split[0] ])
+      boundaries.set(2, [left_right_split[2] + right_top_split[2], 0, right_top_img.columns - right_top_split[2], top_bottom_split[0] ]   )  
+
+      right_bottom_split = Sqed::BoundaryFinder.color_boundary_finder(image: right_bottom_img, scan: :columns, sample_subdivision_size: 2, boundary_color: @boundary_color) # horizontal line b/w (5,3) & 4 
+      
+      bottom_right_top_img = right_bottom_img.crop(0,0, img.columns - left_right_split[2], right_bottom_split[1], true) # 3,5 - we leave right_bottom_split at [1] (not 2) to take into account possible overlap error (crop gets full line at top of image)
+
+      boundaries.set(3, [ left_right_split[2] + right_top_split[2], top_bottom_split[2], left_right_split[2] + right_top_split[2], bottom_right_top_img.rows ] )   
+      boundaries.set(5, [ left_right_split[2], top_bottom_split[2], left_right_split[2] + right_top_split[0], bottom_right_top_img.rows ] )
+      
+      boundaries.set(4, [ left_right_split[2], top_bottom_split[2] + right_top_split[2], img.columns - left_right_split[2],  right_bottom_img.rows - right_top_split[2] ] ) 
+
+    when :vertical_split 
+      t = Sqed::BoundaryFinder.color_boundary_finder(image: img, boundary_color: @boundary_color)  #detect vertical division
       return if t.nil?
       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
 
     when :horizontal_split
-      t = Sqed::BoundaryFinder.color_boundary_finder(image: img, scan: :columns, boundary_color: @boundary_color)  # set to detect horizontal division, (green line)
+      t = Sqed::BoundaryFinder.color_boundary_finder(image: img, scan: :columns, boundary_color: @boundary_color)  # set to detect horizontal division
       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
+    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, use_thumbnail: false ).boundaries
 
       irt = img.crop(*vertical.for(1), true)
@@ -50,7 +75,7 @@ class Sqed::BoundaryFinder::ColorLineFinder < Sqed::BoundaryFinder
       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, use_thumbnail: false).boundaries  # fails
+      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) ]) 

data/lib/sqed/parser/ocr_parser.rb

@@ -61,7 +61,15 @@ class Sqed::Parser::OcrParser < Sqed::Parser
     },
     deterimination_labels: {
       psm: 3
+    },
+    other_labels: {
+      psm: 3
+    },
+    collecting_event_labels: {
+      psm: 3
     }
+
+
   }
 
   # the text extracted from the image
@@ -85,16 +93,18 @@ class Sqed::Parser::OcrParser < Sqed::Parser
   # 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.write('foo4.jpg') 
   # 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('foo5.jpg') 
+  # #img.write('foo3.jpg') 
   #
-  # img.write('foo.jpg') # for debugging purposes, this is the image that is sent to OCR
+  # img.write('foo.jpg') 
   # img = img.white_threshold(245)
   # img
   # end
-  
+ 
+  # @return [String]
+  #   the ocr text 
   def text(section_type: :default)
     img = @image 
     params = SECTION_PARAMS[:default].merge(SECTION_PARAMS[section_type])

data/lib/sqed/version.rb

@@ -1,3 +1,3 @@
 class Sqed
-  VERSION = "0.1.7"
+  VERSION = "0.1.8"
 end

data/lib/sqed_config.rb

@@ -58,6 +58,14 @@ module SqedConfig
   #   | 0 |  :internal_box
   #   -----
   #
+  #    0 | 1 | 2  
+  #   ------------
+  #      | 5 | 3    :seven_slot
+  #    6 |--------
+  #      |   4
+  #
+  #
+  #
 
   # Hash values are used to stub out
   # the Sqed::Boundaries instance.
@@ -69,29 +77,38 @@ module SqedConfig
     vertical_split: [0,1],
     right_t: [0,1,2],
     left_t: [0,1,2],
-    internal_box: [0]
+    internal_box: [0],
+    seven_slot: [0,1,2,3,4,5,6]
   }
 
   # 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
-    :annotated_specimen,    # a specimen is present, and metadata is too
-    :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,
-    :curator_metadata,      # the section contains text with curator metadata
-    :nothing                # section is empty 
+    :annotated_specimen,      # a specimen is present, and metadata is too
+    :collecting_event_labels, # the section that contains collecting event labels (only)
+    :curator_metadata,        # the section contains text with curator metadata
+    :determination_labels,    # the section contains text that determines the specimen (only)
+    :identifier,              # the section contains an identifier (e.g. barcode or unique number)
+    :image_registration,      # the section contains only image registration information,
+    :labels,                  # the section contains collecting event and other non-determination labels
+    :nothing,                 # section is empty 
+    :other_labels,            # the section that contains text that misc. 
+    :specimen,                # the specimen only, no metadata should be present
+    :stage,                   # the image contains the full stage 
   ] 
 
   # Links section types to data parsers
   SECTION_PARSERS = {
-    labels: [ Sqed::Parser::OcrParser ],
-    identifier: [ Sqed::Parser::BarcodeParser, Sqed::Parser::OcrParser ],
-    deterimination_labels: [ Sqed::Parser::OcrParser ],
+    annotated_specimen: [ Sqed::Parser::OcrParser],
+    collecting_event_labels: [ Sqed::Parser::OcrParser],
     curator_metadata: [  Sqed::Parser::OcrParser ],
-    annotated_specimen: [ Sqed::Parser::OcrParser] 
+    deterimination_labels: [ Sqed::Parser::OcrParser ],
+    identifier: [ Sqed::Parser::BarcodeParser, Sqed::Parser::OcrParser ],
+    image_registration: [],
+    labels: [ Sqed::Parser::OcrParser ],
+    nothing: [],
+    other_labels: [ Sqed::Parser::OcrParser ],
+    specimen: [],
+    stage: []
   }
 
   EXTRACTION_PATTERNS = {
@@ -123,6 +140,12 @@ module SqedConfig
       boundary_finder: Sqed::BoundaryFinder::StageFinder,
       layout: :internal_box, 
       metadata_map: {0 => :stage}
+    },
+    
+    seven_slot: {
+      boundary_finder: Sqed::BoundaryFinder::ColorLineFinder,
+      layout: :seven_slot,
+      metadata_map: {0 => :collecting_event_labels, 1 => :determination_labels, 2 => :other_labels, 3 => :image_registration, 4 => :curator_metadata, 5 => :identifier, 6 => :specimen }
     }
   }
 

data/spec/lib/stage_handling/seven_slot_spec.rb

@@ -0,0 +1,18 @@
+require 'spec_helper'
+
+describe 'handling 7 slot stages' do
+
+  let(:image) { ImageHelpers.inhs_stage_7_slot }
+  let(:sqed) { Sqed.new( image: image, pattern: :seven_slot, boundary_color: :red, has_border: false ) }
+
+  context 'parses' do
+    specify 'new() without errors' do
+      expect( sqed ).to be_truthy
+    end
+
+    specify 'get_result without errors' do
+      expect( sqed.result ).to be_truthy
+    end
+  end
+
+end 

data/spec/support/image_helpers.rb

@@ -31,7 +31,9 @@ module ImageHelpers
       get_image 'stage_images/frost_stage_medimum.jpg'
     end
 
-
+    def inhs_stage_7_slot
+      get_image 'stage_images/inhs_7_slot.jpg'
+    end
 
     def crossy_green_line_specimen
       get_image 'stage_images/CrossyGreenLinesSpecimen.jpg'

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: sqed
 version: !ruby/object:Gem::Version
-  version: 0.1.7
+  version: 0.1.8
 platform: ruby
 authors:
 - Matt Yoder
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-09-17 00:00:00.000000000 Z
+date: 2015-12-21 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -176,6 +176,7 @@ files:
 - spec/lib/sqed/parser_spec.rb
 - spec/lib/sqed/result_spec.rb
 - spec/lib/sqed_spec.rb
+- spec/lib/stage_handling/seven_slot_spec.rb
 - spec/spec_helper.rb
 - spec/support/files/barcode_images/code_128_barcode.png
 - spec/support/files/barcode_images/datamatrix_barcode.png
@@ -197,6 +198,8 @@ files:
 - spec/support/files/stage_images/frost_stage_medium.jpg
 - spec/support/files/stage_images/frost_stage_thumb.jpg
 - spec/support/files/stage_images/greenlineimage.jpg
+- spec/support/files/stage_images/inhs_7_slot.jpg
+- spec/support/files/stage_images/inhs_four_thirds.jpg
 - spec/support/files/test0.jpg
 - spec/support/files/test1.jpg
 - spec/support/files/test2.jpg
@@ -242,6 +245,7 @@ test_files:
 - spec/lib/sqed/parser_spec.rb
 - spec/lib/sqed/result_spec.rb
 - spec/lib/sqed_spec.rb
+- spec/lib/stage_handling/seven_slot_spec.rb
 - spec/spec_helper.rb
 - spec/support/files/barcode_images/code_128_barcode.png
 - spec/support/files/barcode_images/datamatrix_barcode.png
@@ -263,6 +267,8 @@ test_files:
 - spec/support/files/stage_images/frost_stage_medium.jpg
 - spec/support/files/stage_images/frost_stage_thumb.jpg
 - spec/support/files/stage_images/greenlineimage.jpg
+- spec/support/files/stage_images/inhs_7_slot.jpg
+- spec/support/files/stage_images/inhs_four_thirds.jpg
 - spec/support/files/test0.jpg
 - spec/support/files/test1.jpg
 - spec/support/files/test2.jpg