Calculating WND-CHARM Features for a grayscale image

In [1]:

    

from wndcharm import diagnostics
print diagnostics


    

    




WND-CHARM Python API Diagnostics
================================
Executable:
	/usr/local/opt/python/bin/python2.7
PYTHONPATH environment variable:
	/Users/colettace/src/wnd-charm/build/lib.macosx-10.10-x86_64-2.7
WND-CHARM library path:
	/Users/colettace/src/wnd-charm/build/lib.macosx-10.10-x86_64-2.7/wndcharm/wndcharm.pyc
Package versions:
	python    	2.7.11 (default, Jan 22 2016, 08:28:37)  [GCC 4.2.1 Compatible Apple LLVM 7.0.2 (clang-700.1.81)]
	wndcharm  	0.9.14+9050371
	numpy     	1.11.1
	scipy     	0.18.0
	matplotlib	1.5.1
	sklearn   	0.17.1
	skimage   	0.11.3
	IPython   	5.0.0
	tifffile  	0.7.0
	PIL       	1.1.7
	pandas    	0.17.1

In [2]:

    

%matplotlib inline


    

In [3]:

    

import numpy as np


    

In [4]:

    

from skimage.io import imshow, imsave


    

In [5]:

    

from skimage.data import chelsea


    

In [6]:

    

imshow( chelsea())


    

    
Out[6]:





<matplotlib.image.AxesImage at 0x110beeb10>






    

In [7]:

    

img = chelsea()


    

In [8]:

    

type( img)


    

    
Out[8]:





numpy.ndarray

In [9]:

    

img.shape


    

    
Out[9]:





(300, 451, 3)

In [10]:

    

img.dtype


    

    
Out[10]:





dtype('uint8')

In [11]:

    

for d, label in enumerate( ('r', 'g', 'b')):
    plane = img[:,:,d]
    print "{}: min={}, max={}".format( label, plane.min(), plane.max())


    

    




r: min=2, max=215
g: min=4, max=189
b: min=0, max=231

In [12]:

    

from skimage.color import rgb2gray


    

In [13]:

    

gray_img = rgb2gray( img)


    

In [14]:

    

gray_img.shape


    

    
Out[14]:





(300, 451)

In [15]:

    

gray_img.dtype


    

    
Out[15]:





dtype('float64')

In [16]:

    

gray_img.min(), gray_img.max()


    

    
Out[16]:





(0.01512078431372549, 0.75561098039215691)

In [17]:

    

gray_img = (gray_img * 255 ).astype( np.uint8 )


    

In [18]:

    

gray_img.dtype


    

    
Out[18]:





dtype('uint8')

In [19]:

    

gray_img.min(), gray_img.max()


    

    
Out[19]:





(3, 192)

In [20]:

    

imshow( gray_img)


    

    
Out[20]:





<matplotlib.image.AxesImage at 0x110e66b90>






    

In [21]:

    

from wndcharm.PyImageMatrix import PyImageMatrix


    

In [22]:

    

matrix = PyImageMatrix()
matrix.allocate(gray_img.shape[1], gray_img.shape[0])
numpy_matrix = matrix.as_ndarray()
numpy_matrix[:] = gray_img


    

In [23]:

    

from wndcharm.FeatureVector import FeatureVector


    

In [24]:

    

fv = FeatureVector( name='FromNumpyMatrix', long=True, original_px_plane=matrix )


    

In [25]:

    

len( fv.values)


    

    




---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-25-7bf0b30980f0> in <module>()
----> 1 len( fv.values)

TypeError: object of type 'NoneType' has no len()

In [26]:

    

import time


    

In [27]:

    

t1 = time.time()
fv.GenerateFeatures(quiet=False, write_to_disk=True)
t2 = time.time()
print "Feat calc took {:0.2f} seconds.".format( t2-t1 )


    

    




CALCULATING FEATURES FROM None <FeatureVector "FromNumpyMatrix" fs_col=0>
CALCULATED: <FeatureVector "FromNumpyMatrix" n_features=2919 fs_col=0>
Writing signature file "FromNumpyMatrix-l.sig"
Feat calc took 12.05 seconds.

In [28]:

    

fv.values


    

    
Out[28]:





array([  4.41400000e+04,   4.81422488e-02,   1.07444423e-01, ...,
        -4.98516383e+01,   2.01940375e+02,   6.98836130e-01])

In [29]:

    

len(fv.values)


    

    
Out[29]:





2919

In [30]:

    

imsave( "chelsea_tiff.tiff", gray_img )


    

In [31]:

    

ls


    

    




FeaturesWithoutTiff.ipynb  chelsea_tiff.tiff
FromNumpyMatrix-l.sig

In [32]:

    

!tiffinfo chelsea_tiff.tiff


    

    




TIFF Directory at offset 0x8 (8)
  Subfile Type: (0 = 0x0)
  Image Width: 451 Image Length: 300
  Bits/Sample: 8
  Sample Format: unsigned integer
  Compression Scheme: None
  Photometric Interpretation: min-is-black
  Samples/Pixel: 1
  Rows/Strip: 300
  Planar Configuration: single image plane
  ImageDescription: {"shape": [300, 451]}
  Software: tifffile.py
  DateTime: 2016:08:31 10:31:54

In [33]:

    

tiff_fv = FeatureVector( name='FromTiff', long=True, source_filepath='chelsea_tiff.tiff' )


    

In [34]:

    

t1 = time.time()
tiff_fv.GenerateFeatures(quiet=False, write_to_disk=True)
t2 = time.time()
print "Feat calc took {:0.2f} seconds.".format( t2-t1 )


    

    




CALCULATING FEATURES FROM chelsea_tiff.tiff <FeatureVector "FromTiff" fs_col=0>
CALCULATED: <FeatureVector "FromTiff" n_features=2919 fs_col=0>
Writing signature file "chelsea_tiff-l.sig"
Feat calc took 11.62 seconds.

In [35]:

    

tiff_fv.values


    

    
Out[35]:





array([  4.41400000e+04,   4.81422488e-02,   1.07444423e-01, ...,
        -4.98516383e+01,   2.01940375e+02,   6.98836130e-01])

In [36]:

    

from wndcharm.utils import compare


    

In [37]:

    

compare( tiff_fv.values, fv.values )


    

    
Out[37]:





True

In [38]:

    

tiff_fv.values == fv.values


    

    
Out[38]:





array([ True,  True,  True, ...,  True,  True,  True], dtype=bool)

In [39]:

    

all( tiff_fv.values == fv.values )


    

    
Out[39]:





True