process.rb

require 'mongo'
require 'progress_bar'
db = Mongo::Connection.new["colors"]
identify = "identify %s"
convert = "convert %s -format %%c -colors 5 -colorspace LAB histogram:info:-"

def run(cmd, *args)
  str = cmd % args
  `#{str}`
end

db["pictures"].ensure_index("path", :unique => true)
db["pictures"].ensure_index([["matches.c1", Mongo::GEO2D]], :min => 0, :max => 255)
db["pictures"].ensure_index([["matches.c2", Mongo::GEO2D]], :min => 0, :max => 255)
db["pictures"].ensure_index([["matches.c3", Mongo::GEO2D]], :min => 0, :max => 255)
db["pictures"].ensure_index([["matches.c4", Mongo::GEO2D]], :min => 0, :max => 255)
db["pictures"].ensure_index([["matches.c5", Mongo::GEO2D]], :min => 0, :max => 255)

files = Dir.glob("in/**/*.jpg")
bar = ProgressBar.new files.length
files.each do |f|
  _, x, y = run(identify, f).match(/(?<x>\d+)x(?<y>\d+)/).to_a
  area = x.to_i * y.to_i
  lines = run convert, f
  matches = {
    c1: [],
    c2: [],
    c3: [],
    c4: [],
    c5: []
  }
  lines.split(/\n/).each do |line|
    _, c, l, a, b = line.match(/(\d+):\s*\(\s*(\d+),\s*(\d+),\s*(\d+)/).to_a.map(&:to_i)
    percent = c / area.to_f
    5.times do |i|
      cutoff = 0.6 - (i * 0.12)
      if percent >= cutoff
        matches[:"c#{i+1}"] << [a, b]
        break
      end
    end
  end
  db["pictures"].insert(path: f, matches: matches)
  bar.increment!
end