downsample.py

# Color of each block at 10, 10
COLORS = (
		(158, 221, 255), # Diamond
		(247, 183, 0), #Yellow
		(1, 185, 1), # Green
		(186, 115, 255), # Purple
		(242, 0, 16), # Red
		(6, 104, 253)) # Blue

def downsample_pixarray(pixarray, factor=40.):
	"""
	Get the nearest color index match from the COLORS tuple
	from every square.

	Diamond dash squares are 40x40.
	"""
	factor = float(factor)

	rows, cols, _ = pixarray.shape

	new_rows = int(math.ceil(rows / factor))
	new_cols = int(math.ceil(cols / factor))

	counts = numpy.zeros((new_rows, new_cols))
	out_pixarray = numpy.zeros((new_rows, new_cols, 3))

	# Downsample instead of averaging
	for i in range(new_rows):
		for j in range(new_cols):
            # Get a point from 10,10 in a 40px square, which should be the right color
			i_ind = int(i * factor + factor/4) 
			j_ind = int(j * factor + factor/4)

			pixel = pixarray[i_ind][j_ind]

			c = nearest_index_to_color(pixel)
			counts[i][j] = c

	return counts

def nearest_index_to_color(color_tuple):
	"Get the index in COLORS closest to color_tuple"
	distances = [color_distance(c, color_tuple) for c in COLORS]
	return distances.index(min(distances))

def normalize_color(color_tuple):
	"Get the color closest to color_tuple"
	return COLORS[nearest_index_to_color(color_tuple)]

def color_distance(c1, c2):
	"""Get the sum of squares of the difference between two colors"""
	return (c1[0] - c2[0]) ** 2 + (c1[1] - c2[1]) ** 2 + (c1[2] - c2[2]) ** 2