scanned page splitter

splitter.py

from PIL import Image
import math, os, sys

def calculate_seam(im, band_width, middle_bias = True, middle_bias_strength = 2):
	pix = im.load()
	size = im.size
	max_std_dev = 0
	max_col = 1
	for i in range(band_width, im.size[0]-band_width+1):
		multiplier = (1-(4*math.sqrt((i-im.size[0]/2)**2)/im.size[0]/2))**middle_bias_strength
		std_dev = 0
		for j in range(0, im.size[1]):
			# calculate sliding band average
			avg_left = 0
			for i2 in range(1, band_width+1):
				avg_left += pix[i-i2, j]
			avg_left /= band_width
			avg_right = 0
			for i2 in range(0, band_width):
				avg_right += pix[i+i2, j]
			avg_right /= band_width
			# calculate the standard deviation between pixel band averages
			std_dev += (avg_right - avg_left)**2
		if middle_bias:
			#print multiplier
			std_dev *= multiplier
		std_dev /= im.size[1]
		if std_dev > max_std_dev:
			max_std_dev = std_dev
			max_col = i
	return max_col, max_std_dev

def split_save(im, fname, split, outdir = "."):
	im1 = im.crop((0, 0, split-1, im.size[1]))
	im2 = im.crop((split, 0, im.size[0], im.size[1]))
	im1.load()
	im2.load()
	fp = fname.split('.')
	im1.save(outdir+"/"+fp[0]+"_01."+fp[1])
	im2.save(outdir+"/"+fp[0]+"_02."+fp[1])

if __name__ == "__main__":
	# test script
	"""fname = "test.jpg"
	im = Image.open(fname)
	seam = calculate_seam(im, 20)
	split_save(im, fname, seam[0])
	"""
	status = True
	
	if(len(sys.argv) > 1):
		input_dir = sys.argv[1]
		output_dir = input_dir+"/"+"split"
		if not os.path.exists(output_dir):
			os.mkdir(output_dir)
		listing = os.listdir(input_dir)
		for infile in listing:
			im = Image.open(input_dir+"/"+infile)
			seam = calculate_seam(im, 20)
			split_save(im, infile, seam[0], outdir=output_dir)
			if status:
				print "[complete] " + infile
	else:
		print "missing command line parameter: input directory"