find_duplicate_images

#!/usr/local/bin/env python

__author__ = 'buchi.busireddy'

from optparse import OptionParser
import hashlib
import os
import time


def main():
    """
    The main method.
    """
    parser = OptionParser()
    parser.add_option("-R", "--root", dest="root", help="Root directory for the images.")
    (options, args) = parser.parse_args()

    if options.root is None:
        print("Please provide the root directory with --root.")
        return

    # Rename all the files with space in the name to have "_" instead of space.
    for root, dirs, filenames in os.walk(options.root):
        for f in filenames:
            file = os.path.join(root, f)

            if " " in file or "(" in file or ")" in file:
                new = file.replace(" ", "_")
                new = new.replace("(", "_")
                new = new.replace(")", "_")
                new = new.replace('__', "_")
                print "Renaming ", file, new
                os.rename(file, new)

    filename_to_hash = {}
    for root, dirs, filenames in os.walk(options.root):
        for f in filenames:
            file = os.path.join(root, f)
            if file.lower().endswith(('.jpg', '.jpeg', '.png')):

                start = time.time()
                image_file = open(file).read()
                hash = hashlib.md5(image_file).hexdigest()

                if hash in filename_to_hash:
                    print filename_to_hash[hash], file
                else:
                    filename_to_hash[hash] = file

if __name__ == "__main__":
    main()