ocarina 0.0.1

data/lib/ocarina/util.rb

@@ -0,0 +1,92 @@
+require 'RMagick'
+
+module Ocarina
+
+  DATA_DIR     = "#{File.dirname(__FILE__)}/../../data"
+  IMAGES_DIR   = "#{DATA_DIR}/images"
+
+  module Util
+
+    def int_to_binary_string(i)
+      i.to_s(2).rjust(config.num_outputs, '0')
+    end
+
+    def char_to_binary_string(target_char)
+      int_to_binary_string target_char.ord
+    end
+
+    # function that maps its input to a range between 0..1
+    # mathematically it's supposed to be asymptotic, but large values of x will round up to 1
+    def sigma(x)
+      1.0/(1.0+Math.exp(-x))
+    end
+
+    def pixel_number_to_col(n, image)
+      n % image.columns
+    end
+
+    def pixel_number_to_row(n, image)
+      n / image.columns
+    end
+
+    def pixel_to_bit(pixel)
+      pixel.red | pixel.green | pixel.black > 0 ? 1 : 0
+    end
+
+
+    # returns a Magick::Image, constructed from the reference gif file for the given character
+    #
+    def reference_image_for_char(char)
+      Magick::Image.read(filename_for_training_image char, "gif").first
+    end
+
+    # returns a Magick::Image, constructed from the noise gif file for the given character
+    #
+    def noise_image_for_char(char)
+      Magick::Image.read(filename_for_noise_image char, "gif").first
+    end
+
+    # returns filepath for the image for a given character
+    #
+    def filename_for_training_image(c, suffix)
+
+      # the default OSX file system is case preserving, meaning that you cannot have
+      # both "A.bitmap" and "a.bitmap" at the same time. So we map lowercase letters to
+      # a special name.
+      #
+      is_lower?(c) ? "#{IMAGES_DIR}/reference/#{c}_lower.#{suffix}" : "#{IMAGES_DIR}/reference/#{c}.#{suffix}"
+    end
+
+    # returns filepath for the "noise" image for a given character
+    #
+    def filename_for_noise_image(c, suffix)
+
+      # the default OSX file system is case preserving, meaning that you cannot have
+      # both "A.bitmap" and "a.bitmap" at the same time. So we map lowercase letters to
+      # a special name.
+      #
+      is_lower?(c) ? "#{IMAGES_DIR}/noise/#{c}_lower.#{suffix}" : "#{IMAGES_DIR}/noise/#{c}.#{suffix}"
+    end
+
+    def filename_for_quantized_image(c, suffix)
+
+      is_lower?(c) ? "#{IMAGES_DIR}/quantized/#{c}_lower.#{suffix}" : "#{IMAGES_DIR}/quantized/#{c}.#{suffix}"
+    end
+
+    # kind of sad that this is not built into Ruby
+    def is_lower?(c)
+      !! /[[:lower:]]/.match(c)
+    end
+
+    def is_upper?(c)
+      !! /[[:upper:]]/.match(c)
+    end
+
+    def quantize_image(image)
+      image.white_threshold(Magick::MaxRGB * 0.35).quantize(2, Magick::GRAYColorspace)
+    end
+
+  end
+
+end
+

data/lib/ocarina/version.rb

@@ -0,0 +1,3 @@
+module Ocarina
+  VERSION = "0.0.1"
+end

data/ocarina.gemspec

@@ -0,0 +1,20 @@
+# -*- encoding: utf-8 -*-
+require File.expand_path('../lib/ocarina/version', __FILE__)
+
+Gem::Specification.new do |gem|
+  gem.authors       = ["George Armhold"]
+  gem.email         = ["armhold@gmail.com"]
+  gem.description   = %q{A Ruby project that uses machine learning to perform Optical Character Recognition}
+  gem.summary       = %q{A Ruby project that uses machine learning to perform Optical Character Recognition}
+  gem.homepage      = "https://github.com/armhold/ocarina"
+
+  gem.files         = `git ls-files`.split($\)
+  gem.executables   = gem.files.grep(%r{^bin/}).map{ |f| File.basename(f) }
+  gem.test_files    = gem.files.grep(%r{^(test|spec|features)/})
+  gem.name          = "ocarina"
+  gem.require_paths = ["lib"]
+  gem.version       = Ocarina::VERSION
+
+  gem.add_runtime_dependency 'rmagick'
+  gem.add_runtime_dependency 'powerbar'
+end

data/test/test_letterpress.rb

@@ -0,0 +1,60 @@
+require 'test/unit'
+require 'ocarina'
+
+class LetterpressTest < Test::Unit::TestCase
+
+
+  def setup
+    @network = Ocarina::Network.load_network_from_file "#{Ocarina::DATA_DIR}/letterpress-train.bin"
+    @cropper = Ocarina::LetterpressCropper.new(@network.config)
+  end
+
+  def test_boards
+
+    assert_equal [
+                     %w(P R B R Z),
+                     %w(T A V Z R),
+                     %w(B D A K Y),
+                     %w(G I G K F),
+                     %w(R Y S J V)
+                 ], @cropper.decipher_board(@network, read_board("board1.png"))
+
+    assert_equal [
+                     %w(Q D F P M),
+                     %w(N E E S I),
+                     %w(A W F M L),
+                     %w(F R P T T),
+                     %w(K C S S Y)
+                 ], @cropper.decipher_board(@network, read_board("board2.png"))
+
+    assert_equal [
+                     %w(L H F L M),
+                     %w(R V P U K),
+                     %w(V O E E X),
+                     %w(I N R I T),
+                     %w(V N S I Q)
+                 ], @cropper.decipher_board(@network, read_board("board3.png"))
+
+    assert_equal [
+                     %w(C U H S Z),
+                     %w(N T V K I),
+                     %w(E X K O S),
+                     %w(B C U P N),
+                     %w(W E E V E)
+                 ], @cropper.decipher_board(@network, read_board("board4.png"))
+
+    assert_equal [
+                     %w(C Y M T I),
+                     %w(P Z Y L Y),
+                     %w(D W O H S),
+                     %w(D W H A S),
+                     %w(O Z X G K)
+                 ], @cropper.decipher_board(@network, read_board("board5.png"))
+  end
+
+  def read_board(file)
+    Magick::Image.read("#{Ocarina::DATA_DIR}/images/letterpress/#{file}").first
+  end
+
+
+end

metadata

@@ -0,0 +1,231 @@
+--- !ruby/object:Gem::Specification
+name: ocarina
+version: !ruby/object:Gem::Version
+  version: 0.0.1
+  prerelease: 
+platform: ruby
+authors:
+- George Armhold
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2012-11-01 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: rmagick
+  requirement: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: powerbar
+  requirement: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+description: A Ruby project that uses machine learning to perform Optical Character
+  Recognition
+email:
+- armhold@gmail.com
+executables: []
+extensions: []
+extra_rdoc_files: []
+files:
+- .gitignore
+- Gemfile
+- LICENSE
+- NOTICE
+- README.md
+- Rakefile
+- data/.gitkeep
+- data/images/letterpress/board1.png
+- data/images/letterpress/board2.png
+- data/images/letterpress/board3.png
+- data/images/letterpress/board4.png
+- data/images/letterpress/board5.png
+- data/images/noise/.gitkeep
+- data/images/noise/0.gif
+- data/images/noise/1.gif
+- data/images/noise/2.gif
+- data/images/noise/3.gif
+- data/images/noise/4.gif
+- data/images/noise/5.gif
+- data/images/noise/6.gif
+- data/images/noise/7.gif
+- data/images/noise/8.gif
+- data/images/noise/9.gif
+- data/images/noise/A.gif
+- data/images/noise/B.gif
+- data/images/noise/C.gif
+- data/images/noise/D.gif
+- data/images/noise/E.gif
+- data/images/noise/F.gif
+- data/images/noise/G.gif
+- data/images/noise/H.gif
+- data/images/noise/I.gif
+- data/images/noise/J.gif
+- data/images/noise/K.gif
+- data/images/noise/L.gif
+- data/images/noise/M.gif
+- data/images/noise/N.gif
+- data/images/noise/O.gif
+- data/images/noise/P.gif
+- data/images/noise/Q.gif
+- data/images/noise/R.gif
+- data/images/noise/S.gif
+- data/images/noise/T.gif
+- data/images/noise/U.gif
+- data/images/noise/V.gif
+- data/images/noise/W.gif
+- data/images/noise/X.gif
+- data/images/noise/Y.gif
+- data/images/noise/Z.gif
+- data/images/noise/a_lower.gif
+- data/images/noise/b_lower.gif
+- data/images/noise/c_lower.gif
+- data/images/noise/d_lower.gif
+- data/images/noise/e_lower.gif
+- data/images/noise/f_lower.gif
+- data/images/noise/g_lower.gif
+- data/images/noise/h_lower.gif
+- data/images/noise/i_lower.gif
+- data/images/noise/j_lower.gif
+- data/images/noise/k_lower.gif
+- data/images/noise/l_lower.gif
+- data/images/noise/m_lower.gif
+- data/images/noise/n_lower.gif
+- data/images/noise/o_lower.gif
+- data/images/noise/p_lower.gif
+- data/images/noise/q_lower.gif
+- data/images/noise/r_lower.gif
+- data/images/noise/s_lower.gif
+- data/images/noise/t_lower.gif
+- data/images/noise/u_lower.gif
+- data/images/noise/v_lower.gif
+- data/images/noise/w_lower.gif
+- data/images/noise/x_lower.gif
+- data/images/noise/y_lower.gif
+- data/images/noise/z_lower.gif
+- data/images/reference/.gitkeep
+- data/images/reference/0.gif
+- data/images/reference/1.gif
+- data/images/reference/2.gif
+- data/images/reference/3.gif
+- data/images/reference/4.gif
+- data/images/reference/5.gif
+- data/images/reference/6.gif
+- data/images/reference/7.gif
+- data/images/reference/8.gif
+- data/images/reference/9.gif
+- data/images/reference/A.gif
+- data/images/reference/B.gif
+- data/images/reference/C.gif
+- data/images/reference/D.gif
+- data/images/reference/E.gif
+- data/images/reference/F.gif
+- data/images/reference/G.gif
+- data/images/reference/H.gif
+- data/images/reference/I.gif
+- data/images/reference/J.gif
+- data/images/reference/K.gif
+- data/images/reference/L.gif
+- data/images/reference/M.gif
+- data/images/reference/N.gif
+- data/images/reference/O.gif
+- data/images/reference/P.gif
+- data/images/reference/Q.gif
+- data/images/reference/R.gif
+- data/images/reference/S.gif
+- data/images/reference/T.gif
+- data/images/reference/U.gif
+- data/images/reference/V.gif
+- data/images/reference/W.gif
+- data/images/reference/X.gif
+- data/images/reference/Y.gif
+- data/images/reference/Z.gif
+- data/images/reference/a_lower.gif
+- data/images/reference/b_lower.gif
+- data/images/reference/c_lower.gif
+- data/images/reference/d_lower.gif
+- data/images/reference/e_lower.gif
+- data/images/reference/f_lower.gif
+- data/images/reference/g_lower.gif
+- data/images/reference/h_lower.gif
+- data/images/reference/i_lower.gif
+- data/images/reference/j_lower.gif
+- data/images/reference/k_lower.gif
+- data/images/reference/l_lower.gif
+- data/images/reference/m_lower.gif
+- data/images/reference/n_lower.gif
+- data/images/reference/o_lower.gif
+- data/images/reference/p_lower.gif
+- data/images/reference/q_lower.gif
+- data/images/reference/r_lower.gif
+- data/images/reference/s_lower.gif
+- data/images/reference/t_lower.gif
+- data/images/reference/u_lower.gif
+- data/images/reference/v_lower.gif
+- data/images/reference/w_lower.gif
+- data/images/reference/x_lower.gif
+- data/images/reference/y_lower.gif
+- data/images/reference/z_lower.gif
+- data/letterpress-train.bin
+- data/train.bin
+- lib/ocarina.rb
+- lib/ocarina/character_generator.rb
+- lib/ocarina/config.rb
+- lib/ocarina/error_stats.rb
+- lib/ocarina/letterpress_character_generator.rb
+- lib/ocarina/letterpress_cropper.rb
+- lib/ocarina/network.rb
+- lib/ocarina/util.rb
+- lib/ocarina/version.rb
+- ocarina.gemspec
+- ocarina.png
+- test/test_letterpress.rb
+homepage: https://github.com/armhold/ocarina
+licenses: []
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubyforge_project: 
+rubygems_version: 1.8.24
+signing_key: 
+specification_version: 3
+summary: A Ruby project that uses machine learning to perform Optical Character Recognition
+test_files:
+- test/test_letterpress.rb