performance - The fastest way to work with ndimage in python? -

i have function iterates on ndimage (converts image 1 color space another). works slow (2 cores cpu, 2.3 ghz, image size = 3 mp):

1) brute-force approach (loop): 27 seconds

def imagergbtoycrcb(rgb_image):     """ converts image rgb ycrcb. overwrites.      """     w, h = rgb_image.shape[0], rgb_image.shape[1]      y in range(h):         x in range(w):             rgb = rgb_image[x][y]             ycrcb = rgbtoycrcb(rgb)             rgb_image[x][y] = ycrcb              return rgb_image  def rgbtoycrcb(rgb):     """ converts rgb vector ycrcb vector      keyword arguments:     rgb -- list of size 3: [r,g,b]      returns: ycrcr color (list of size 3: [y,cr,cb])     """      r,g,b = float(rgb[0]),float(rgb[1]),float(rgb[2])     y = 0.299*r + 0.587*g + 0.114*b     cb = 128 - 0.1687*r - 0.3313*g + 0.5*b      cr = 128 + 0.5*r - 0.4187*g - 0.0813*b      return [y,cr,cb] 

2) vectorization approach (numpy.apply_along_axis, numpy.dot): 90 seconds (???)

import numpy np  matr_to_ycrcb_mult = np.array([ [0.299, 0.587, 0.114], [0.5, -0.4187, -0.0813], [-0.1687, -0.3313, 0.5] ]) vec_to_ycrcb_add = np.array([ 0, 128, 128 ]) matr_to_rgb_mult = np.array([ [1,1.402,0], [1,-0.71414,-0.34414], [1,0,1.772] ]) vec_to_rgb_add = np.array([ -128*1.402, 128*1.05828, -128*1.772 ])  def imagergbtoycrcb(rgb_image):     rgb_image = np.apply_along_axis(rgbtoycrcb, 2, rgb_image)     return rgb_image  def rgbtoycrcb(rgb):     """ converts rgb vector ycrcb vector     """     return np.dot(matr_to_ycrcb_mult,rgb) + vec_to_ycrcb_add 

is there faster way process ndimage?
have implemented vectorization concepts correctly?

apply_along_axis convenience function not vectorize operation, iteration still takes place in python land.

your rgbtoycrcb function needs tiny change operate whole image. in function new_rgbtoycrcb2 slice out r, g, b (these views original image data, no copying) , stack them @ end.

you can reshape , use dot product without slicing , glueing, may faster (function new_rgbtoycrcb2):

import numpy np skimage.data import coffee  def imagergbtoycrcb(rgb_image):     """ converts image rgb ycrcb. overwrites.      """     w, h = rgb_image.shape[0], rgb_image.shape[1]      y in range(h):         x in range(w):             rgb = rgb_image[x][y]             ycrcb = rgbtoycrcb(rgb)             rgb_image[x][y] = ycrcb              return rgb_image  def rgbtoycrcb(rgb):     """ converts rgb vector ycrcb vector      keyword arguments:     rgb -- list of size 3: [r,g,b]      returns: ycrcr color (list of size 3: [y,cr,cb])     """      r,g,b = float(rgb[0]),float(rgb[1]),float(rgb[2])     y = 0.299*r + 0.587*g + 0.114*b     cb = 128 - 0.1687*r - 0.3313*g + 0.5*b      cr = 128 + 0.5*r - 0.4187*g - 0.0813*b      return [y,cr,cb]  def new_rgbtoycrcb(image):     r, g, b = image[..., 0], image[..., 1], image[..., 2]      y = 0.299*r + 0.587*g + 0.114*b     cb = 128 - 0.1687*r - 0.3313*g + 0.5*b      cr = 128 + 0.5*r - 0.4187*g - 0.0813*b      return np.dstack((y, cr, cb))  def new_rgbtoycrcb2(image):     m = np.array([[0.299, 0.587, 0.114],                   [0.5, - 0.4187, - 0.0813],                   [-0.1687, - 0.3313, 0.5]])     result = m.dot(image.reshape(-1, 3).t).t     result[..., 1:] += 128.0     return result.reshape(image.shape)  image = coffee() / 255.0 result = new_rgbtoycrcb(image) print(np.allclose(result, imagergbtoycrcb(image)))