In [4]:

    

from PIL import *
import numpy as np
from pylab import *
import cv2
import numpy as num,math
from math import pi,sqrt,exp
from copy import deepcopy
import matplotlib.pyplot as plt
from numpy import matrix
import matplotlib.cm as cm
%matplotlib inline


    

In [5]:

    

I1 = array(Image.open('/home/srikar/CVV/susan_input1.png'))
imshow(I1,cmap = cm.gray)
S1 = np.shape(I1)


    

In [14]:

    

#salt and pepper noise 
I2 = array(Image.open('/home/srikar/CVV/susan_input2.png'))
imshow(I2,cmap = cm.gray)
S2 =np.shape(I2)


    

In [21]:

    

#the header files
from numpy import *
import numpy as np
from PIL import Image
import matplotlib.pyplot as plt
import matplotlib.cm as cm
import math
import os,sys
import scipy.ndimage
import time
import scipy




def Susan_corner(I1, S1):
    mask_radius = 3
    length = 7 #2*3+1
    maximum = 0
    susan_mask = np.ones((7,7))
    #copying the image into I2
    I1[:] = I2[:]
    R = np.zeros(S1)#stores the distances
    x_min = y_min = mask_radius
    x_max = S1[0] - mask_radius
    y_max = S1[1] - mask_radius

    for i in range(x_min, x_max):
        for j in range(y_min, y_max):
            susan_mask[:] = I1[i-mask_radius:i+mask_radius+1,j-mask_radius:j+mask_radius+1]   #incremeting the individual index starting from the radius
            centre = mask_radius  #the nucleus position
            intensity = susan_mask[mask_radius, mask_radius]  #the intensity of the nucleus pixel

            for m in range(0,length):
                for n in range(0,length):
                    if ((m-centre)*(m-centre)+(n-centre)*(n-centre)<=mask_radius*mask_radius):
                        susan_mask[m][n]=math.exp(-math.pow(((susan_mask[m][n]-intensity)/27),6))#n(r0) formula as given by equation 0.5 and threshold value is found by trial and error method
                    else:
                        susan_mask[m][n] = 0
            susan_mask[mask_radius, mask_radius] = 0
            n_pixel = np.sum(susan_mask)
            R[i][j] = n_pixel
            if maximum < n_pixel:
                maximum = n_pixel
    print (maximum)
    print (np.sum(R))


    #to threshold the matrix USAN
    g = maximum/2
    for i in range(x_min,x_max):
        for j in range(y_min,y_max):
            if(R[i][j] >=g):
                R[i][j] = 0
            else:
                R[i][j] = g - R[i][j]

    print(np.sum(np.nonzero(R)))
    plt.figure()
    plt.imshow(R, cmap = cm.gray)
    plt.show()

    Ix = np.array(np.matrix(R))
    Iy = np.array(np.matrix(R))
    O = np.ones((S1[0], S1[1]), np.float32)
    for i in range(S1[0]):
        for j in range(S1[1]):
            O[i,j] = math.degrees(math.atan2(Iy[i][j], Ix[i][j]))+180

    plt.figure(2)
    plt.imshow(O, cmap = cm.gray)
    plt.show() #showing the raw plotted corners
Image1 = Image.open("/home/srikar/CVV/susan_input1.png").convert('L')
Image2 = Image.open("/home/srikar/CVV/susan_input2.png").convert('L')
I2 = np.array(Image1)
S1 = np.shape(I2)
I1 = np.zeros(S1)
I3 = np.array(Image2)
S2 = np.shape(I3)
I4 = np.zeros(S2)
Susan_corner(I1, S1)
Susan_corner(I4, S2)


    

    




28.0
1671998.50756
50845






    













    













    




28.0
1671998.50756
50845






    













    

In [15]:

    




    

In [16]:

    




    

In [ ]:

    




    

In [ ]:

    




    

In [ ]:

    




    

In [ ]:

    




    

In [ ]:

    




    

In [9]:

    

I1 = array(Image.open('/home/srikar/CVV/susan_input1.png'))
imshow(I1,cmap = cm.gray)
S1 = np.shape(I1)


    

In [10]:

    

#salt and pepper noise 
I2 = array(Image.open('/home/srikar/CVV/susan_input2.png'))
imshow(I2,cmap = cm.gray)
S2 =np.shape(I2)


    

In [ ]: