lib/sqed/parser/ocr_parser.rb in sqed-0.1.7 vs lib/sqed/parser/ocr_parser.rb in sqed-0.1.8

- old
+ new

@@ -59,11 +59,19 @@ labels: { psm: 3, # may need to be 6 }, deterimination_labels: { psm: 3 + }, + other_labels: { + psm: 3 + }, + collecting_event_labels: { + psm: 3 } + + } # the text extracted from the image attr_accessor :text @@ -83,19 +91,21 @@ # 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.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]) r = RTesseract.new(img, params) @text = r.to_s.strip