Add clean script.

examples/clean.rb

@@ -0,0 +1,77 @@
+#!/usr/bin/env ruby
+
+#This script is used to convert a bunch of grayscale image to black&white and
+#then compile the in a pdf.
+
+SCRIPT_DIR = File.dirname($0)
+$LOAD_PATH << SCRIPT_DIR
+$LOAD_PATH << "#{SCRIPT_DIR}/../lib"
+
+require 'rubygems'
+require 'leptonica'
+require 'score_tools'
+
+#Convert a grayscale image to a b&w image of size w x h.
+def clean_image(image, w, h)
+    image = ScoreTools.deskew(image)
+    image = ScoreTools.adaptive_map_threshold(image)
+    image = ScoreTools.resize(image, w, h)
+    image = ScoreTools.remove_edge_noise(image)
+    image = ScoreTools.center!(image)
+    image
+end
+
+out_file_prefix = "out"
+page = '0000'
+double_page = false
+name = 'out'
+width, height = 0, 0
+
+i = 0
+while(ARGV[i] =~ /^-/)
+    if(ARGV[i] == '-double-page')
+        double_page = true
+        i += 1
+    elsif(ARGV[i] == '-name')
+        i += 1
+        name = ARGV[i]
+        i += 1
+    elsif(ARGV[i] == '-size')
+        i += 1
+        width, height = ARGV[i].split('x').map{|x|x.to_i}
+        i += 1
+    end
+end
+
+if(width == 0 || height==0)
+    puts "Invalid size"
+    exit
+end
+
+files = ARGV[i..-1]
+files.each do |file|
+    puts page
+    image = Leptonica::Pix.read(file)
+
+    if(double_page)
+        left, right = ScoreTools.split_in_half(image)
+
+        left = clean_image(left, width, height)
+        left.write("#{out_file_prefix}#{page}.tiff", :tiff_g4)
+        page.succ!
+
+        right = clean_image(right, width, height)
+        right.write("#{out_file_prefix}#{page}.tiff", :tiff_g4)
+        page.succ!
+    else
+        image = clean_image(image, width, height)
+        image.write("#{out_file_prefix}#{page}.tiff", :tiff_g4)
+        page.succ!
+    end
+
+    #Free the image (hopefully)
+    GC.start
+end
+
+ScoreTools.files_to_pdf(Dir["out*.tiff"].sort, name)
+

examples/find_pattern.rb

@@ -1,8 +1,7 @@
 #!/usr/bin/env ruby
 
-#To run without installing the gem uncomment the following lines:
-#SCRIPT_DIR = File.dirname($0)
-#$LOAD_PATH << "#{SCRIPT_DIR}/../lib"
+SCRIPT_DIR = File.dirname($0)
+$LOAD_PATH << "#{SCRIPT_DIR}/../lib"
 
 require 'rubygems'
 require 'leptonica'

examples/score_tools.rb

@@ -0,0 +1,108 @@
+require 'leptonica'
+
+module ScoreTools
+    def self.adaptive_map_threshold(pix)
+        reduction = 4
+        map = pix.background_norm_gray_morph(reduction, 11, 128)
+        norm = pix.apply_inv_background_gray_map(map, reduction)
+        threshold = norm.estimate_global_threshold
+        norm.threshold(threshold)
+    end
+
+    def self.deskew(pix)
+        skew = 0.0
+        if(pix.depth == 1)
+            skew = pix.find_skew
+        else
+            threshold = pix.estimate_global_threshold
+            skew = pix.threshold(threshold).find_skew
+        end
+        pix.rotate(degrees_to_radians(skew))
+    end
+
+    def self.split_in_half(pix)
+        w = pix.width
+        h = pix.height
+        lbox = Leptonica::Box.create(0, 0, w/2, h)
+        rbox = Leptonica::Box.create(w/2+1, 0, w/2, h)
+        [pix.clip(lbox), pix.clip(rbox)]
+    end
+
+    def self.resize(pix, new_w, new_h)
+        w = pix.width
+        h = pix.height
+        w_diff = new_w - w
+        h_diff = new_h - h
+
+        w_add = w_diff > 0 ? w_diff : 0
+        h_add = h_diff > 0 ? h_diff : 0
+        pix_expanded = if(w_add > 0 || h_add > 0)
+                           pix.add_border(0, w_add, 0, h_add)
+                       else
+                           pix
+                       end
+
+        w_sub = w_diff < 0 ? w_diff.abs : 0
+        h_sub = h_diff < 0 ? h_diff.abs : 0
+        pix_reduced = if(w_sub > 0 || h_sub > 0)
+                          pix_expanded.remove_border(0, w_sub, 0, h_sub)
+                      else
+                          pix_expanded
+                      end
+    end
+
+    def self.files_to_pdf(files, filename)
+        `tiffcp #{files.join(' ')} #{filename}.tiff`
+        `tiff2pdf #{filename}.tiff -t"#{filename.gsub('_', ' ')}" -z -o #{filename}.pdf`
+        `rm #{filename}.tiff`
+    end
+
+    def self.remove_edge_noise(image, smudge_factor = 51, max_iterations = 4)
+        #remove some initial noise
+        sel_b = Leptonica::Sel.create_brick(3, 3, 1, 1)
+
+        #smudging bricks
+        smudge_center = smudge_factor/2
+        sel_h = Leptonica::Sel.create_brick(1, smudge_factor, 1, smudge_center)
+        sel_v = Leptonica::Sel.create_brick(smudge_factor, 1, smudge_center, 1)
+
+        count = 0
+        content_mask = image.open(sel_b)
+        loop do
+            content_mask.dilate!(sel_h)
+            content_mask.dilate!(sel_v)
+            #BUG: Sometimes the far right column doesn't get set by a
+            #horizontal dilate. Setting the border pixels as a workaround.
+            content_mask.set_border!(1, 1, 1, 1)
+            content_mask = content_mask.remove_border_components
+            #content_mask.write("mask#{count}.tiff", :tiff_g4)
+            count += 1
+            #p content_mask.count_connected_components
+            break if(content_mask.count_connected_components <= 1 || count >= max_iterations)
+        end
+        image.and(content_mask)
+    end
+
+    def self.center!(image)
+        #Do an opening to hopefully avoid some noise. This might remove some
+        #content. If it does open with a smaller brick.
+        sel = Leptonica::Sel.create_brick(7, 7, 3, 3)
+        boxa = image.open(sel).connected_components
+        bounding_box = boxa.extent
+        x = bounding_box.x
+        y = bounding_box.y
+        w = bounding_box.w
+        h = bounding_box.h
+        sx = (image.width - w - 2*x)/2
+        sy = (image.height - h - 2*y)/2
+        image.shift!(sx, sy)
+    end
+
+    def self.degrees_to_radians(angle)
+        angle*(Math::PI/180.0)
+    end
+
+    def self.radians_to_degrees(angle)
+        angle*(180.0/Math::PI)
+    end
+end