Slow Python - what can be done?
Peter Otten
__peter__ at web.de
Sat Mar 20 20:33:48 CET 2004
Jason wrote:
> My uber-abstraction is due to recently fleeing from C (abstraction
> made hard) and Lisp (no workable GUI stuff) so I'm stuck in the
> middle. Anyway, I appreciate the comments.
Not so fast. I think abstraction is *good* - only not well suited for the
inner loop of an image transformation algorithm.
Your warping seems to boil down to an affine transform, so whatever fancy
stuff you do as a preparation - with points, lines and all that nice
object-oriented stuff - you always end up with two equations
u = ax + by + c
v = dx + ey + f
and that's the only thing you should calculate repeatedly if you want
efficiency.
Using that recipe reduced calculation time for a small 350x223 pixel image
from 10 to 0.33 (0.2 with psyco) seconds. Here's the code, and I'm
confident you'll recognize it :-)
(no testing performed, but the images *do* look similar)
<warp.py>
""" call with
--psyco to use psyco
--old to use the original algorithm
an image file as the *last* parameter
"""
from Tkinter import *
import Image
import ImageTk
from sys import exit
from math import sqrt
if "--psyco" in sys.argv:
import psyco
psyco.full()
class Point:
# A Point in the plane
def __init__(self, int1, int2):
# Constructor
self.x = float(int1)
self.y = float(int2)
def __add__(self, other):
# Add two points
return Point(self.x + other.x, self.y + other.y)
def __sub__(self, other):
# Sub two points
return Point(self.x - other.x, self.y - other.y)
def __mul__(self, other):
# Either mult by a constant or dot product
if type(other) == float or type(other) == int:
return Point(self.x*other, self.y*other)
else:
return self.x*other.x + self.y*other.y
def __div__(self,other):
# division by a constant
if type(other) == float or type(other) == int:
return Point(self.x/other, self.y/other)
def __rmul__(self, other):
# multiplication by a constant
return Point(self.x*other, self.y*other)
def __rdiv__(self, other):
# division by a constant
return Point(other/self.x, other/self.y)
def __str__(self):
# printing represenation
return '(%s, %s)' % (self.x, self.y)
def length(self):
# regular length
return sqrt(pow(self.x, 2) + pow(self.y, 2))
def perpindicular(self):
# 90 deg rotation
return Point(self.y, -self.x)
def to_tuple(self):
# makes a tuple of ints
return (int(self.x), int(self.y))
class WarpLine:
# The lines used to warp the image
def __init__(self, x0, y0, x1, y1, id):
# Constructor - just two points - id not used yet.
self.id = 0
self.point1 = Point(x0, y0)
self.point2 = Point(x1, y1)
def __str__(self):
# Printing
return '%s->%s' % (self.point1, self.point2)
def length(self):
# Segment length
return sqrt(pow(self.point2.x-self.point1.x, 2) +
pow(self.point2.y-self.point1.y, 2))
def getUV(self, point):
# v = shortest distance of point to line
# u = the parameterization of the closest point from v
diff = (self.point2 - self.point1)
u = ((point - self.point1) * diff) / (diff * diff)
v = ((point - self.point1) * diff.perpindicular()) / sqrt(diff *
diff)
return u, v
def transformPoint(self, line, point):
# finds transform of point based on self and line
diff = (line.point2 - line.point1)
u, v = self.getUV(point)
return line.point1 + u * diff + (v * diff.perpindicular())
/sqrt(diff * diff)
class Picture:
# A simple image class
def __init__(self, file):
# Load up an image
self.data = Image.open(file)
def in_bounds(self, pt):
# is point in our bounds?
if pt.x < 0 or pt.y < 0 or pt.x > self.data.size[0] - 1 or pt.y >
self.data.size[1] - 1:
return 0
else:
return 1
def coefficients(self, transform=None):
orig = transform(Point(0, 0))
p = transform(Point(1, 0)) - orig
q = transform(Point(0, 1)) - orig
a, b, c = p.x, q.x, orig.x
d, e, f = p.y, q.y, orig.y
return a, b, c, d, e, f
def warp_new(self, source, line1, line2):
""" psyco doesn't like lambdas, so I had to factor it out.
Does anybody know why?
"""
self._warp(source,
*self.coefficients(lambda p: line1.transformPoint(line2, p)))
def _warp(self, source, a, b, c, d, e, f):
width, height = self.data.size
dest = [0] * (width*height)
src = source.data.getdata()
yoff = 0
for y in range(height):
for x in range(width):
u = int(a*x+b*y+c)
v = int(d*x+e*y+f)
if u >= 0 and u < width and v >= 0 and v < height:
dest[x + yoff] = src[u + v*width]
yoff += width
self.data.putdata(dest)
def warp_old(self, source, line1, line2):
# Do transformPoint on each pixel, save results
# This is the slow part of the program
dest = list(self.data.getdata())
src = source.data.getdata()
for x in range(0, self.data.size[0] - 1):
for y in range(0, self.data.size[1] - 1):
xy = line1.transformPoint(line2,Point(x,y)).to_tuple()
if self.in_bounds(Point(xy[0], xy[1])):
dest[x + y*self.data.size[0]] = src[xy[0] +
xy[1]*self.data.size[0]]
else:
dest[x + y*self.data.size[0]] = 0
self.data.putdata(dest)
def show(self):
# show the image
root = Tk()
canvas = Canvas(root,
width=self.data.size[0],height=self.data.size[1])
canvas.pack()
photo = ImageTk.PhotoImage(self.data)
disp = canvas.create_image(0, 0, anchor=NW, image=photo)
mainloop()
if __name__ == "__main__":
import time
p1 = Picture(sys.argv[-1])
line1 = WarpLine(0, 0, 200, 50, None)
line2 = WarpLine(-100, 0, 150, 0, None)
start = time.time()
if "--old" in sys.argv:
p1.warp_old(p1, line1, line2)
else:
p1.warp_new(p1, line1, line2)
print time.time() - start
p1.show()
</warp.py>
I'm sure there's room for improvement. E. g., you could devise a clipping
algorithm to not calculate all the black points. By the way, the Python
Imaging Library (PIL) has such a transform built in - but that might spoil
the fun.
Peter
More information about the Python-list
mailing list