quantize.py

import numpy as np
from bisect import bisect_right
import argparse, os

# TODO: try frequency analysis quantization
# e.g. k-means? from http://scikit-learn.org/stable/auto_examples/cluster/plot_lena_compress.html

# These need to both be numpy arrays
def quantize(data, levels):
  indices = levels.searchsorted(data, side="right")
  levels = np.append([0], levels)
  return levels[indices]

parser = argparse.ArgumentParser(description='Quantize LCS vector to compress better.')
parser.add_argument('filename')
mode = parser.add_mutually_exclusive_group(required=True)
mode.add_argument('-q', dest='levels', metavar='N', nargs='+',
                    help='a quantization level.', type=int)
mode.add_argument('-t', dest='threshold', metavar='<threshold>', help='remove all below this minimum value.', type=int)

args = parser.parse_args()
filename, levels, threshold = args.filename, args.levels, args.threshold

# Add check for file
if not os.path.exists(filename):
  print "Error: file %s does not exist" % (filename)

lcs = np.fromfile(filename, dtype=np.uint8)
#max_val = lcs.max()

if threshold:
  print "Removing values below %s." % (threshold)
  lcs[lcs < threshold] = 0
  output = lcs
if levels:
  # if 0 not in levels: levels = [0]+sorted(levels)
  print "Quantizing into these buckets: %s" % (levels)
  levels = np.array(levels)
  output = quantize(lcs, levels)

output.tofile(filename+".qzd")