In [1]:

    

from skimage import io, color, feature, transform
import numpy as np
from skimage.filters import rank
import plotly.plotly as py
import cufflinks as cf # this is necessary to link pandas to plotly
import plotly.graph_objs as go
cf.go_offline()
from plotly.offline import plot, iplot
from skimage.morphology import disk
from matplotlib import pyplot as PLT


    

In [2]:

    

rgbImg = io.imread("/Users/sreejithmenon/Dropbox/Social_Media_Wildlife_Census/All_Zebra_Count_Images/8598.jpeg")


    

In [ ]:

    

hsvImg = color.rgb2hsv(rgbImg)


    

In [ ]:

    

red = np.array([pix[0] for row in rgbImg for pix in row])
green = np.array([pix[1] for row in rgbImg for pix in row])
blue = np.array([pix[2] for row in rgbImg for pix in row])

hue = np.array([pix[0] for row in hsvImg for pix in row])
saturation = np.array([pix[1] for row in hsvImg for pix in row])
value = np.array([pix[2] for row in hsvImg for pix in row])


    

In [ ]:

    

y = 0.299 * red + 0.587 * green + 0.114 * blue
(np.max(y) - np.min(y))/np.mean(y)


    

In [ ]:

    

a = np.array(rgbImg)


    

In [ ]:

    

a


    

In [ ]:

    

hue


    

In [ ]:

    

data = [
    go.Histogram(
    x = hue,
    nbinsx = 12
    )
]

plot(data)


    

In [ ]:

    

a, _ = np.histogram(saturation, bins=5)
np.std(a)


    

In [3]:

    

gray = color.rgb2gray(rgbImg)


    

In [ ]:

    

gray.shape


    

In [ ]:

    

glcm = feature.greycomatrix(gray, [1], [0, np.pi/4, np.pi/2, 3*np.pi/4])
glcm.shape


    

In [ ]:

    

entropy = rank.entropy(gray, disk(5))


    

In [ ]:

    

entropy.shape


    

In [ ]:

    

feature.hog(resized_img, orientations=8)


    

In [4]:

    

resized_img = transform.resize(gray, (600,600))


    

In [ ]:

    

resized_img.shape


    

In [ ]:

    

PLT.imshow(flipped)
PLT.show()


    

In [5]:

    

left = resized_img.transpose()[:300].transpose()
right = resized_img.transpose()[300:].transpose()


    

In [ ]:

    

left.shape, right.shape


    

In [15]:

    

I = np.identity(600)[::-1]


    

In [16]:

    

flipped = I.dot(right)


    

In [17]:

    

inner = feature.hog(left) - feature.hog(flipped)


    

In [18]:

    

np.linalg.norm(inner)


    

    
Out[18]:





7.4599010338301044

In [13]:

    

inner


    

    
Out[13]:





array([-0.00700603, -0.00796074, -0.00081415, ...,  0.00238143,
        0.0120759 , -0.00081522])

In [ ]: