croptoelie 0.3.1 → 0.4.0

data/README.md

@@ -6,8 +6,7 @@ Crops images based on entropy: leaving the most interesting part intact.
 
 Don't expect this to be a replacement for human cropping, it is an algorythm and not an extremely smart one at that :).
 
-Best results achieved in combination with scaling: the cropping is then only used to square the image, cutting off the least interesting part. It offers two methods, scanning and trimming: with scanning the whole image is placed in an array then evaluated: very slow and memory-gobbling. 
-
+Best results achieved in combination with scaling: the cropping is then only used to square the image, cutting off the least interesting part. 
 The trimming simply chops off te edge that is least interesting, and continues doing so, untill it reached the requested size.
 
 ## Usage
@@ -44,6 +43,14 @@ File *uploaders/attachement_uploader.rb*:
 * Commit and push until you are happy with your contribution
 * Please try not to mess with the Rakefile, version, or history. If you want to have your own version, or is otherwise necessary, that is fine, but please isolate to its own commit so I can cherry-pick around it.
 
+## Changelog
+2011-04-19: Replace crop with crop! avoids copying large chunks of images around.
+2011-04-18: Limit to N steps, instead of step_size.
+2011-04-16: Introduce tests and a profiler script, to profile performance.
+
+## Todo
+Improved algorythm: first @image.scale by F, investigate the entropy on that, most-interesting square by factor F is to-be-cropped area.
+
 ## Copyright
 
 Copyright (c) 2011 Bèr Kessels. See LICENSE.txt for

data/Rakefile

@@ -18,7 +18,7 @@ Jeweler::Tasks.new do |gem|
   gem.homepage = "http://github.com/berkes/croptoelie"
   gem.license = "MIT"
   gem.summary = %Q{Content aware cropper.}
-  gem.description = %Q{Crops images based on entropy: leaving the most interesting part intact. Don't expect this to be a replacement for human cropping, it is an algorythm and not an extremely smart one at that :). Best results achieved in combination with scaling: the cropping is then only used to square the image, cutting off the least interesting part. It offers two methods, scanning and trimming: with scanning the whole image is placed in an array then evaluated: very slow and memory-gobbling. The trimming simply chops off te edge that is least interesting, and continues doing so, untill it reached the requested size.}
+  gem.description = %Q{Crops images based on entropy: leaving the most interesting part intact. Don't expect this to be a replacement for human cropping, it is an algorythm and not an extremely smart one at that :). Best results achieved in combination with scaling: the cropping is then only used to square the image, cutting off the least interesting part. The trimming simply chops off te edge that is least interesting, and continues doing so, untill it reached the requested size.}
   gem.email = "ber@webschuur.com"
   gem.authors = ["Bèr Kessels"]
   # Include your dependencies below. Runtime dependencies are required when using your gem,

data/VERSION

@@ -1 +1 @@
-0.3.0
+0.4.0

data/croptoelie.gemspec

@@ -5,12 +5,12 @@
 
 Gem::Specification.new do |s|
   s.name = %q{croptoelie}
-  s.version = "0.3.1"
+  s.version = "0.4.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Bèr Kessels"]
-  s.date = %q{2011-04-07}
-  s.description = %q{Crops images based on entropy: leaving the most interesting part intact. Don't expect this to be a replacement for human cropping, it is an algorythm and not an extremely smart one at that :). Best results achieved in combination with scaling: the cropping is then only used to square the image, cutting off the least interesting part. It offers two methods, scanning and trimming: with scanning the whole image is placed in an array then evaluated: very slow and memory-gobbling. The trimming simply chops off te edge that is least interesting, and continues doing so, untill it reached the requested size.}
+  s.date = %q{2011-04-19}
+  s.description = %q{Crops images based on entropy: leaving the most interesting part intact. Don't expect this to be a replacement for human cropping, it is an algorythm and not an extremely smart one at that :). Best results achieved in combination with scaling: the cropping is then only used to square the image, cutting off the least interesting part. The trimming simply chops off te edge that is least interesting, and continues doing so, untill it reached the requested size.}
   s.email = %q{ber@webschuur.com}
   s.extra_rdoc_files = [
     "LICENSE.txt",
@@ -30,7 +30,10 @@ Gem::Specification.new do |s|
     "doc/croptoelie_test.rb",
     "doc/histogram.rb",
     "lib/croptoelie.rb",
+    "test/entropyish.png",
+    "test/entropyish.txt",
     "test/helper.rb",
+    "test/profiler.rb",
     "test/test_croptoelie.rb"
   ]
   s.homepage = %q{http://github.com/berkes/croptoelie}
@@ -40,6 +43,7 @@ Gem::Specification.new do |s|
   s.summary = %q{Content aware cropper.}
   s.test_files = [
     "test/helper.rb",
+    "test/profiler.rb",
     "test/test_croptoelie.rb"
   ]
 

data/lib/croptoelie.rb

@@ -3,16 +3,15 @@ class CropToelie
   include Magick
   
   attr_accessor :orig
-  attr_accessor :step_size
+  attr_accessor :steps
   
   # Create a new CropToelie object from a ImageList single image object. 
   #  If you want to provide a file by its path use CropToelie.from_file('/path/to/image.png'). 
   def initialize(image)
     @image = image
-    @orig = image
     
     # Hardcoded (but overridable) defaults.
-    @step_size  = 10
+    @steps  = 10
 
     # Preprocess image.
     @image = @image.quantize
@@ -28,32 +27,21 @@ class CropToelie
     return CropToelie.new(image)
   end
 
-  # Crops an image to width x height
-  # 
-  # If you want speed, choose :by_trim. It will use smart_crop_by_trim().
-  # If you have a very crowded image, an image with lots of color, or dark and light
-  #  spots, the much slower :by_search will give better results. It will use
-  #  smart_crop_by_search().  
-  def smart_crop(width, height, method = :by_trim)
-    sq = square(width, height, method)
-    return @orig.crop(sq[:left], sq[:top], width, height, true)
+  # Crops an image to width x height 
+  def smart_crop(width, height)
+    sq = square(width, height)
+    return @image.crop!(sq[:left], sq[:top], width, height, true)
   end
 
   # Squares an image (with smart_square) and then scales that to width, heigh
-  # 
-  # If you want speed, choose :by_trim. It will use smart_crop_by_trim().
-  # If you have a very crowded image, an image with lots of color, or dark and light
-  #  spots, the much slower :by_search will give better results. It will use
-  #  smart_crop_by_search().  
   def smart_crop_and_scale(width, height)
-    cropped = smart_square    
-    return cropped.scale(width, height)
+    smart_square    
+    return @image.scale!(width, height)
   end
   
   # Squares an image by slicing off the least interesting parts. 
   # Usefull for squaring images such as thumbnails. Usefull before scaling.
   def smart_square
-    cropped = @orig  #square images can be returned as-is.
     if @rows != @columns #None-square images must be shaved off.
       if @rows < @columns #landscape
         crop_height = crop_width = @rows
@@ -61,23 +49,18 @@ class CropToelie
         crop_height = crop_width = @columns
       end
 
-      sq = square(crop_width, crop_height, :by_trim)
-      cropped = @orig.crop(sq[:left], sq[:top], crop_width, crop_height, true)
+      sq = square(crop_width, crop_height)
+      @image.crop!(sq[:left], sq[:top], crop_width, crop_height, true)
     end
     
-    cropped    
+    @image    
   end
   
   # Finds the most interesting square with size width x height.
   # 
-  # See smart_crop documentation for explanation about the method
   # Returns a hash {:left => left, :top => top, :right => right, :bottom => bottom}
-  def square(width, height, method = :by_trim)
-    if method == :by_trim
-      return smart_crop_by_trim(width, height)
-    else
-      return smart_crop_by_search(width, height)
-    end
+  def square(width, height)
+    return smart_crop_by_trim(width, height)
   end
   
   private
@@ -88,49 +71,15 @@ class CropToelie
       return (@columns > @width) && (@rows < @height)
     end
 
-    # Find Entropy by moving the "to be cropped" area over the image and 
-    # recording the entropy of each such square. 
-    # The square with the highest entropy is considered the most interesting and
-    # cropped out of the original.
-    # NOTE: this method is very slow compared to smart_crop_by_trim. 
-    def smart_crop_by_search(requested_x, requested_y)
-      left, top     = 0, 0
-      right, bottom = requested_x, requested_y
-
-      # Create a hash with all entropies
-      entropies = {}
-              
-      # start in left-top corner, walk to right, with steps of 10 px.
-      while (bottom <= @rows)
-        while (right <= @columns)
-          square = {:left => left, :top => top, :right => right, :bottom => bottom}
-          entropies[square] = entropy_slice(@image, left, top, right - left, bottom - top)
-          
-          left += @step_size
-          right += @step_size
-        end
-        # @TODO last item is the one that goes over the edge, or touches the edge.
-        left = 0
-        right = requested_x
-        top += @step_size
-        bottom += @step_size
-      end
-
-      # Find the square with highest entropy
-      best = entropies.max_by{|s| s[1]}[0]
-      
-      # chop that one out
-      best
-    end
-  
     def smart_crop_by_trim(requested_x, requested_y)
       left, top = 0, 0
       right, bottom = @columns, @rows
       width, height = right, bottom
+      step_size = step_size(requested_x, requested_y)
 
       # Slice from left and right edges until the correct width is reached.
       while (width > requested_x)
-        slice_width = [(width - requested_x), @step_size].min
+        slice_width = [(width - requested_x), step_size].min
 
         left_entropy  = entropy_slice(@image, left, 0, slice_width, bottom)
         right_entropy = entropy_slice(@image, (right - slice_width), 0, slice_width, bottom)
@@ -147,7 +96,7 @@ class CropToelie
 
       # Slice from top and bottom edges until the correct height is reached.
       while (height > requested_y)
-        slice_height = [(height - @step_size), @step_size].min
+        slice_height = [(height - step_size), step_size].min
         
         top_entropy    = entropy_slice(@image, 0, top, @columns, slice_height)
         bottom_entropy = entropy_slice(@image, 0, (bottom - slice_height), @columns, slice_height)
@@ -172,6 +121,8 @@ class CropToelie
     end
     
     # Compute the entropy of an image, defined as -sum(p.*log2(p)).
+    #  Note: instead of log2, only available in ruby > 1.9, we use
+    #  log(p)/log(2). which has the same effect.
     def entropy(image_slice)
       hist = image_slice.color_histogram
       hist_size = hist.values.inject{|sum,x| sum ? sum + x : x }.to_f
@@ -179,9 +130,12 @@ class CropToelie
       entropy = 0
       hist.values.each do |h| 
         p = h.to_f / hist_size
-        entropy += (p * Math.log2(p)) if p != 0
+        entropy += (p * (Math.log(p)/Math.log(2))) if p != 0
       end
-     
       return entropy * -1
     end
+    
+    def step_size(requested_x, requested_y)
+      ((([@rows - requested_x, @columns - requested_y].max)/2)/@steps).to_i
+    end
 end

data/test/helper.rb

@@ -1,5 +1,6 @@
 require 'rubygems'
 require 'bundler'
+
 begin
   Bundler.setup(:default, :development)
 rescue Bundler::BundlerError => e
@@ -14,5 +15,3 @@ $LOAD_PATH.unshift(File.join(File.dirname(__FILE__), '..', 'lib'))
 $LOAD_PATH.unshift(File.dirname(__FILE__))
 require 'croptoelie'
 
-class Test::Unit::TestCase
-end

data/test/profiler.rb

@@ -0,0 +1,37 @@
+require 'ruby-prof'
+$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), '..', 'lib'))
+$LOAD_PATH.unshift(File.dirname(__FILE__))
+require 'croptoelie'
+
+tests = {
+  :smart_crop_by_trim => {:method => :smart_crop, :params => [100, 100]}, 
+#  :smart_crop_by_search => {:method => :smart_crop, :params => [100, 100, :by_search]},
+  :smart_crop_and_scale => {:method => :smart_crop_and_scale, :params => [100, 100]},
+  :smart_square => {:method => :smart_square, :params => []}
+}
+#result = RubyProf.profile do
+  tests.each do |id, test|
+    filename = File.join(File.expand_path(File.dirname(__FILE__)), "../doc/tyto.jpg")
+    
+    2.times do |i|
+#      result = RubyProf.profile do
+        img = CropToelie.from_file(filename)
+        img.send(test[:method], *test[:params])
+        img = nil
+#      end
+
+  #    # Print a flat profile to text
+  #    puts "Run #{i}:\t #{id} ------------------------"
+  #    #file = File.new("./#{id}-#{i}.txt", "w")
+  #    printer = RubyProf::FlatPrinter.new(result)
+  #    printer.print(STDOUT, {:min_percent => 10})
+  #    print = nil
+    end
+  end
+#end
+#    # Print a flat profile to text
+#    puts "Run #{i}:\t #{id} ------------------------"
+#    #file = File.new("./#{id}-#{i}.txt", "w")
+#    printer = RubyProf::FlatPrinter.new(result)
+#    printer.print(STDOUT, {:min_percent => 10})
+#    print = nil

data/test/test_croptoelie.rb

@@ -1,7 +1,42 @@
 require 'helper'
 
 class TestCroptoelie < Test::Unit::TestCase
-  should "probably rename this file and start testing for real" do
-    flunk "hey buddy, you should probably rename this file and start testing for real"
+  def setup
+    @filename = File.join(File.expand_path(File.dirname(__FILE__)), "entropyish.png")
+    @image    = Magick::ImageList.new(@filename).last
+  end
+  should "initialize a croptoelie image from an ImageList item" do
+    img = CropToelie.new(@image)
+    assert_equal(img.class, CropToelie)
+  end
+  should "create a croptoelie from an imagefile" do
+    img = CropToelie.from_file(@filename)
+    assert_equal(img.class, CropToelie)
+  end
+
+  should "fail on creating a croptoelie image from a textfile" do
+    assert_raise Magick::ImageMagickError, NoMethodError do
+       CropToelie.new(File.join(File.expand_path(File.dirname(__FILE__)), "entropyish.txt"))
+    end
+  end
+  
+  should "crop to 100x100 without scaling with smart_crop" do
+    img = CropToelie.new(@image)
+    img = img.smart_crop(100, 100)
+    size = [img.rows, img.columns]
+    assert_equal(size, [100, 100])
+  end
+  
+  should "crop to 100x100 with scaling with smart_crop_and_scale" do
+    img = CropToelie.new(@image)
+    img = img.smart_crop_and_scale(100, 100)
+    size = [img.rows, img.columns]
+    assert_equal(size, [100, 100])
+  end
+  
+  should "square image without scaling" do
+    img = CropToelie.new(@image)
+    img = img.smart_square
+    assert_equal(img.rows, img.columns)
   end
 end

metadata

@@ -2,7 +2,7 @@
 name: croptoelie
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 0.3.1
+  version: 0.4.0
 platform: ruby
 authors: 
 - "B\xC3\xA8r Kessels"
@@ -10,7 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-04-07 00:00:00 +02:00
+date: 2011-04-19 00:00:00 +02:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -79,7 +79,7 @@ dependencies:
   type: :runtime
   prerelease: false
   version_requirements: *id006
-description: "Crops images based on entropy: leaving the most interesting part intact. Don't expect this to be a replacement for human cropping, it is an algorythm and not an extremely smart one at that :). Best results achieved in combination with scaling: the cropping is then only used to square the image, cutting off the least interesting part. It offers two methods, scanning and trimming: with scanning the whole image is placed in an array then evaluated: very slow and memory-gobbling. The trimming simply chops off te edge that is least interesting, and continues doing so, untill it reached the requested size."
+description: "Crops images based on entropy: leaving the most interesting part intact. Don't expect this to be a replacement for human cropping, it is an algorythm and not an extremely smart one at that :). Best results achieved in combination with scaling: the cropping is then only used to square the image, cutting off the least interesting part. The trimming simply chops off te edge that is least interesting, and continues doing so, untill it reached the requested size."
 email: ber@webschuur.com
 executables: []
 
@@ -102,7 +102,10 @@ files:
 - doc/croptoelie_test.rb
 - doc/histogram.rb
 - lib/croptoelie.rb
+- test/entropyish.png
+- test/entropyish.txt
 - test/helper.rb
+- test/profiler.rb
 - test/test_croptoelie.rb
 has_rdoc: true
 homepage: http://github.com/berkes/croptoelie
@@ -118,7 +121,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
-      hash: -1061105859
+      hash: -831961247
       segments: 
       - 0
       version: "0"
@@ -137,4 +140,5 @@ specification_version: 3
 summary: Content aware cropper.
 test_files: 
 - test/helper.rb
+- test/profiler.rb
 - test/test_croptoelie.rb