Rank filters

Rank filters are non-linear filters using the local greylevels ordering to compute the filtered value. This ensemble of filters share a common base: the local grey-level histogram extraction computed on the neighborhood of a pixel (defined by a 2D structuring element). If the filtered value is taken as the middle value of the histogram, we get the classical median filter.

Rank filters can be used for several purposes such as:

  • image quality enhancement e.g. image smoothing, sharpening
  • image pre-processing e.g. noise reduction, contrast enhancement
  • feature extraction e.g. border detection, isolated point detection
  • post-processing e.g. small object removal, object grouping, contour smoothing

Some well known filters are specific cases of rank filters [1]_ e.g. morphological dilation, morphological erosion, median filters.

The different implementation availables in skimage are compared.

In this example, we will see how to filter a greylevel image using some of the linear and non-linear filters availables in skimage. We use the camera image from skimage.data.

[1]Pierre Soille, On morphological operators based on rank filters, Pattern Recognition 35 (2002) 527-535.
import numpy as np
import matplotlib.pyplot as plt

from skimage import data

ima = data.camera()
hist = np.histogram(ima, bins=np.arange(0, 256))

plt.figure(figsize=(8, 3))
plt.subplot(1, 2, 1)
plt.imshow(ima, cmap=plt.cm.gray, interpolation='nearest')
plt.axis('off')
plt.subplot(1, 2, 2)
plt.plot(hist[1][:-1], hist[0], lw=2)
plt.title('histogram of grey values')
../../_images/plot_rank_filters_1.png

Noise removal

Some noise is added to the image, 1% of pixels are randomly set to 255, 1% are randomly set to 0. The median filter is applied to remove the noise.

Note

there are different implementations of median filter : skimage.filter.median_filter and skimage.filter.rank.median

noise = np.random.random(ima.shape)
nima = data.camera()
nima[noise > 0.99] = 255
nima[noise < 0.01] = 0

from skimage.filter.rank import median
from skimage.morphology import disk

fig = plt.figure(figsize=[10, 7])

lo = median(nima, disk(1))
hi = median(nima, disk(5))
ext = median(nima, disk(20))
plt.subplot(2, 2, 1)
plt.imshow(nima, cmap=plt.cm.gray, vmin=0, vmax=255)
plt.xlabel('noised image')
plt.subplot(2, 2, 2)
plt.imshow(lo, cmap=plt.cm.gray, vmin=0, vmax=255)
plt.xlabel('median $r=1$')
plt.subplot(2, 2, 3)
plt.imshow(hi, cmap=plt.cm.gray, vmin=0, vmax=255)
plt.xlabel('median $r=5$')
plt.subplot(2, 2, 4)
plt.imshow(ext, cmap=plt.cm.gray, vmin=0, vmax=255)
plt.xlabel('median $r=20$')
../../_images/plot_rank_filters_2.png

The added noise is efficiently removed, as the image defaults are small (1 pixel wide), a small filter radius is sufficient. As the radius is increasing, objects with a bigger size are filtered as well, such as the camera tripod. The median filter is commonly used for noise removal because borders are preserved.

Image smoothing

The example hereunder shows how a local mean smoothes the camera man image.

from skimage.filter.rank import mean

fig = plt.figure(figsize=[10, 7])

loc_mean = mean(nima, disk(10))
plt.subplot(1, 2, 1)
plt.imshow(ima, cmap=plt.cm.gray, vmin=0, vmax=255)
plt.xlabel('original')
plt.subplot(1, 2, 2)
plt.imshow(loc_mean, cmap=plt.cm.gray, vmin=0, vmax=255)
plt.xlabel('local mean $r=10$')
../../_images/plot_rank_filters_3.png

One may be interested in smoothing an image while preserving important borders (median filters already achieved this), here we use the bilateral filter that restricts the local neighborhood to pixel having a greylevel similar to the central one.

Note

a different implementation is available for color images in skimage.filter.denoise_bilateral.

from skimage.filter.rank import bilateral_mean

ima = data.camera()
selem = disk(10)

bilat = bilateral_mean(ima.astype(np.uint16), disk(20), s0=10, s1=10)

# display results
fig = plt.figure(figsize=[10, 7])
plt.subplot(2, 2, 1)
plt.imshow(ima, cmap=plt.cm.gray)
plt.xlabel('original')
plt.subplot(2, 2, 3)
plt.imshow(bilat, cmap=plt.cm.gray)
plt.xlabel('bilateral mean')
plt.subplot(2, 2, 2)
plt.imshow(ima[200:350, 350:450], cmap=plt.cm.gray)
plt.subplot(2, 2, 4)
plt.imshow(bilat[200:350, 350:450], cmap=plt.cm.gray)
../../_images/plot_rank_filters_4.png

One can see that the large continuous part of the image (e.g. sky) is smoothed whereas other details are preserved.

Contrast enhancement

We compare here how the global histogram equalization is applied locally.

The equalized image [2] has a roughly linear cumulative distribution function for each pixel neighborhood. The local version [3] of the histogram equalization emphasizes every local greylevel variations.

[2]http://en.wikipedia.org/wiki/Histogram_equalization
[3]http://en.wikipedia.org/wiki/Adaptive_histogram_equalization
from skimage import exposure
from skimage.filter import rank

ima = data.camera()
# equalize globally and locally
glob = exposure.equalize(ima) * 255
loc = rank.equalize(ima, disk(20))

# extract histogram for each image
hist = np.histogram(ima, bins=np.arange(0, 256))
glob_hist = np.histogram(glob, bins=np.arange(0, 256))
loc_hist = np.histogram(loc, bins=np.arange(0, 256))

plt.figure(figsize=(10, 10))
plt.subplot(321)
plt.imshow(ima, cmap=plt.cm.gray, interpolation='nearest')
plt.axis('off')
plt.subplot(322)
plt.plot(hist[1][:-1], hist[0], lw=2)
plt.title('histogram of grey values')
plt.subplot(323)
plt.imshow(glob, cmap=plt.cm.gray, interpolation='nearest')
plt.axis('off')
plt.subplot(324)
plt.plot(glob_hist[1][:-1], glob_hist[0], lw=2)
plt.title('histogram of grey values')
plt.subplot(325)
plt.imshow(loc, cmap=plt.cm.gray, interpolation='nearest')
plt.axis('off')
plt.subplot(326)
plt.plot(loc_hist[1][:-1], loc_hist[0], lw=2)
plt.title('histogram of grey values')
../../_images/plot_rank_filters_5.png

another way to maximize the number of greylevels used for an image is to apply a local autoleveling, i.e. here a pixel greylevel is proportionally remapped between local minimum and local maximum.

The following example shows how local autolevel enhances the camara man picture.

from skimage.filter.rank import autolevel

ima = data.camera()
selem = disk(10)

auto = autolevel(ima.astype(np.uint16), disk(20))

# display results
fig = plt.figure(figsize=[10, 7])
plt.subplot(1, 2, 1)
plt.imshow(ima, cmap=plt.cm.gray)
plt.xlabel('original')
plt.subplot(1, 2, 2)
plt.imshow(auto, cmap=plt.cm.gray)
plt.xlabel('local autolevel')
../../_images/plot_rank_filters_6.png

This filter is very sensitive to local outlayers, see the little white spot in the sky left part. This is due to a local maximum which is very high comparing to the rest of the neighborhood. One can moderate this using the percentile version of the autolevel filter which uses given percentiles (one inferior, one superior) in place of local minimum and maximum. The example below illustrates how the percentile parameters influence the local autolevel result.

from skimage.filter.rank import percentile_autolevel

image = data.camera()

selem = disk(20)
loc_autolevel = autolevel(image, selem=selem)
loc_perc_autolevel0 = percentile_autolevel(image, selem=selem, p0=.00, p1=1.0)
loc_perc_autolevel1 = percentile_autolevel(image, selem=selem, p0=.01, p1=.99)
loc_perc_autolevel2 = percentile_autolevel(image, selem=selem, p0=.05, p1=.95)
loc_perc_autolevel3 = percentile_autolevel(image, selem=selem, p0=.1, p1=.9)

fig, axes = plt.subplots(nrows=3, figsize=(7, 8))
ax0, ax1, ax2 = axes
plt.gray()

ax0.imshow(np.hstack((image, loc_autolevel)))
ax0.set_title('original / autolevel')

ax1.imshow(
    np.hstack((loc_perc_autolevel0, loc_perc_autolevel1)), vmin=0, vmax=255)
ax1.set_title('percentile autolevel 0%,1%')
ax2.imshow(
    np.hstack((loc_perc_autolevel2, loc_perc_autolevel3)), vmin=0, vmax=255)
ax2.set_title('percentile autolevel 5% and 10%')

for ax in axes:
    ax.axis('off')
../../_images/plot_rank_filters_7.png

The morphological contrast enhancement filter replaces the central pixel by the local maximum if the original pixel value is closest to local maximum, otherwise by the minimum local.

from skimage.filter.rank import morph_contr_enh

ima = data.camera()

enh = morph_contr_enh(ima, disk(5))

# display results
fig = plt.figure(figsize=[10, 7])
plt.subplot(2, 2, 1)
plt.imshow(ima, cmap=plt.cm.gray)
plt.xlabel('original')
plt.subplot(2, 2, 3)
plt.imshow(enh, cmap=plt.cm.gray)
plt.xlabel('local morphlogical contrast enhancement')
plt.subplot(2, 2, 2)
plt.imshow(ima[200:350, 350:450], cmap=plt.cm.gray)
plt.subplot(2, 2, 4)
plt.imshow(enh[200:350, 350:450], cmap=plt.cm.gray)
../../_images/plot_rank_filters_8.png

The percentile version of the local morphological contrast enhancement uses percentile p0 and p1 instead of the local minimum and maximum.

from skimage.filter.rank import percentile_morph_contr_enh

ima = data.camera()

penh = percentile_morph_contr_enh(ima, disk(5), p0=.1, p1=.9)

# display results
fig = plt.figure(figsize=[10, 7])
plt.subplot(2, 2, 1)
plt.imshow(ima, cmap=plt.cm.gray)
plt.xlabel('original')
plt.subplot(2, 2, 3)
plt.imshow(penh, cmap=plt.cm.gray)
plt.xlabel('local percentile morphlogical\n contrast enhancement')
plt.subplot(2, 2, 2)
plt.imshow(ima[200:350, 350:450], cmap=plt.cm.gray)
plt.subplot(2, 2, 4)
plt.imshow(penh[200:350, 350:450], cmap=plt.cm.gray)
../../_images/plot_rank_filters_9.png

Image threshold

The Otsu’s threshold [1]_ method can be applied locally using the local greylevel distribution. In the example below, for each pixel, an “optimal” threshold is determined by maximizing the variance between two classes of pixels of the local neighborhood defined by a structuring element.

The example compares the local threshold with the global threshold skimage.filter.threshold_otsu.

Note

Local thresholding is much slower than global one. There exists a function for global Otsu thresholding: skimage.filter.threshold_otsu.

[1]http://en.wikipedia.org/wiki/Otsu’s_method
from skimage.filter.rank import otsu
from skimage.filter import threshold_otsu

p8 = data.page()

radius = 10
selem = disk(radius)

# t_loc_otsu is an image
t_loc_otsu = otsu(p8, selem)
loc_otsu = p8 >= t_loc_otsu

# t_glob_otsu is a scalar
t_glob_otsu = threshold_otsu(p8)
glob_otsu = p8 >= t_glob_otsu

plt.figure()
plt.subplot(2, 2, 1)
plt.imshow(p8, cmap=plt.cm.gray)
plt.xlabel('original')
plt.colorbar()
plt.subplot(2, 2, 2)
plt.imshow(t_loc_otsu, cmap=plt.cm.gray)
plt.xlabel('local Otsu ($radius=%d$)' % radius)
plt.colorbar()
plt.subplot(2, 2, 3)
plt.imshow(p8 >= t_loc_otsu, cmap=plt.cm.gray)
plt.xlabel('original>=local Otsu' % t_glob_otsu)
plt.subplot(2, 2, 4)
plt.imshow(glob_otsu, cmap=plt.cm.gray)
plt.xlabel('global Otsu ($t=%d$)' % t_glob_otsu)
../../_images/plot_rank_filters_10.png

The following example shows how local Otsu’s threshold handles a global level shift applied to a synthetic image .

n = 100
theta = np.linspace(0, 10 * np.pi, n)
x = np.sin(theta)
m = (np.tile(x, (n, 1)) * np.linspace(0.1, 1, n) * 128 + 128).astype(np.uint8)

radius = 10
t = rank.otsu(m, disk(radius))
plt.figure()
plt.subplot(1, 2, 1)
plt.imshow(m)
plt.xlabel('original')
plt.subplot(1, 2, 2)
plt.imshow(m >= t, interpolation='nearest')
plt.xlabel('local Otsu ($radius=%d$)' % radius)
../../_images/plot_rank_filters_11.png

Image morphology

Local maximum and local minimum are the base operators for greylevel morphology.

Note

skimage.dilate and skimage.erode are equivalent filters (see below for comparison).

Here is an example of the classical morphological greylevel filters: opening, closing and morphological gradient.

from skimage.filter.rank import maximum, minimum, gradient

ima = data.camera()

closing = maximum(minimum(ima, disk(5)), disk(5))
opening = minimum(maximum(ima, disk(5)), disk(5))
grad = gradient(ima, disk(5))

# display results
fig = plt.figure(figsize=[10, 7])
plt.subplot(2, 2, 1)
plt.imshow(ima, cmap=plt.cm.gray)
plt.xlabel('original')
plt.subplot(2, 2, 2)
plt.imshow(closing, cmap=plt.cm.gray)
plt.xlabel('greylevel closing')
plt.subplot(2, 2, 3)
plt.imshow(opening, cmap=plt.cm.gray)
plt.xlabel('greylevel opening')
plt.subplot(2, 2, 4)
plt.imshow(grad, cmap=plt.cm.gray)
plt.xlabel('morphological gradient')
../../_images/plot_rank_filters_12.png

Feature extraction

Local histogram can be exploited to compute local entropy, which is related to the local image complexity. Entropy is computed using base 2 logarithm i.e. the filter returns the minimum number of bits needed to encode local greylevel distribution.

skimage.rank.entropy returns local entropy on a given structuring element. The following example shows this filter applied on 8- and 16- bit images.

Note

to better use the available image bit, the function returns 10x entropy for 8-bit images and 1000x entropy for 16-bit images.

from skimage import data
from skimage.filter.rank import entropy
from skimage.morphology import disk
import numpy as np
import matplotlib.pyplot as plt

# defining a 8- and a 16-bit test images
a8 = data.camera()
a16 = data.camera().astype(np.uint16) * 4

ent8 = entropy(a8, disk(5))  # pixel value contain 10x the local entropy
ent16 = entropy(a16, disk(5))  # pixel value contain 1000x the local entropy

# display results
plt.figure(figsize=(10, 10))

plt.subplot(2, 2, 1)
plt.imshow(a8, cmap=plt.cm.gray)
plt.xlabel('8-bit image')
plt.colorbar()

plt.subplot(2, 2, 2)
plt.imshow(ent8, cmap=plt.cm.jet)
plt.xlabel('entropy*10')
plt.colorbar()

plt.subplot(2, 2, 3)
plt.imshow(a16, cmap=plt.cm.gray)
plt.xlabel('16-bit image')
plt.colorbar()

plt.subplot(2, 2, 4)
plt.imshow(ent16, cmap=plt.cm.jet)
plt.xlabel('entropy*1000')
plt.colorbar()
../../_images/plot_rank_filters_13.png

Implementation

The central part of the skimage.rank filters is build on a sliding window that update local greylevel histogram. This approach limits the algorithm complexity to O(n) where n is the number of image pixels. The complexity is also limited with respect to the structuring element size.

from time import time

from scipy.ndimage.filters import percentile_filter
from skimage.morphology import dilation
from skimage.filter import median_filter
from skimage.filter.rank import median, maximum


def exec_and_timeit(func):
    """Decorator that returns both function results and execution time."""
    def wrapper(*arg):
        t1 = time()
        res = func(*arg)
        t2 = time()
        ms = (t2 - t1) * 1000.0
        return (res, ms)
    return wrapper


@exec_and_timeit
def cr_med(image, selem):
    return median(image=image, selem=selem)


@exec_and_timeit
def cr_max(image, selem):
    return maximum(image=image, selem=selem)


@exec_and_timeit
def cm_dil(image, selem):
    return dilation(image=image, selem=selem)


@exec_and_timeit
def ctmf_med(image, radius):
    return median_filter(image=image, radius=radius)


@exec_and_timeit
def ndi_med(image, n):
    return percentile_filter(image, 50, size=n * 2 - 1)

Comparison between

  • rank.maximum
  • cmorph.dilate

on increasing structuring element size

a = data.camera()

rec = []
e_range = range(1, 20, 2)
for r in e_range:
    elem = disk(r + 1)
    rc, ms_rc = cr_max(a, elem)
    rcm, ms_rcm = cm_dil(a, elem)
    rec.append((ms_rc, ms_rcm))

rec = np.asarray(rec)

plt.figure()
plt.title('increasing element size')
plt.ylabel('time (ms)')
plt.xlabel('element radius')
plt.plot(e_range, rec)
plt.legend(['crank.maximum', 'cmorph.dilate'])

and increasing image size

../../_images/plot_rank_filters_14.png
r = 9
elem = disk(r + 1)

rec = []
s_range = range(100, 1000, 100)
for s in s_range:
    a = (np.random.random((s, s)) * 256).astype('uint8')
    (rc, ms_rc) = cr_max(a, elem)
    (rcm, ms_rcm) = cm_dil(a, elem)
    rec.append((ms_rc, ms_rcm))

rec = np.asarray(rec)

plt.figure()
plt.title('increasing image size')
plt.ylabel('time (ms)')
plt.xlabel('image size')
plt.plot(s_range, rec)
plt.legend(['crank.maximum', 'cmorph.dilate'])
../../_images/plot_rank_filters_15.png

Comparison between:

  • rank.median
  • ctmf.median_filter
  • ndimage.percentile

on increasing structuring element size

a = data.camera()

rec = []
e_range = range(2, 30, 4)
for r in e_range:
    elem = disk(r + 1)
    rc, ms_rc = cr_med(a, elem)
    rctmf, ms_rctmf = ctmf_med(a, r)
    rndi, ms_ndi = ndi_med(a, r)
    rec.append((ms_rc, ms_rctmf, ms_ndi))

rec = np.asarray(rec)

plt.figure()
plt.title('increasing element size')
plt.plot(e_range, rec)
plt.legend(['rank.median', 'ctmf.median_filter', 'ndimage.percentile'])
plt.ylabel('time (ms)')
plt.xlabel('element radius')
../../_images/plot_rank_filters_16.png

comparison of outcome of the three methods

plt.figure()
plt.imshow(np.hstack((rc, rctmf, rndi)))
plt.xlabel('rank.median vs ctmf.median_filter vs ndimage.percentile')
../../_images/plot_rank_filters_17.png

and increasing image size

r = 9
elem = disk(r + 1)

rec = []
s_range = [100, 200, 500, 1000]
for s in s_range:
    a = (np.random.random((s, s)) * 256).astype('uint8')
    (rc, ms_rc) = cr_med(a, elem)
    rctmf, ms_rctmf = ctmf_med(a, r)
    rndi, ms_ndi = ndi_med(a, r)
    rec.append((ms_rc, ms_rctmf, ms_ndi))

rec = np.asarray(rec)

plt.figure()
plt.title('increasing image size')
plt.plot(s_range, rec)
plt.legend(['rank.median', 'ctmf.median_filter', 'ndimage.percentile'])
plt.ylabel('time (ms)')
plt.xlabel('image size')
../../_images/plot_rank_filters_18.png
plt.show()

Python source code: download (generated using skimage 0.8.2)