How can I filter the gaussian noise from an image using segmentation?

Image Analysis
#1

Hello guys, I’ve added some Gaussian Noise to my image but it looks like it’s not completely eliminated from my image, what am I doing wrong? This is my code:

# Import numpy, skimage and matplotlib
import numpy as np
from skimage import io,color, util
import matplotlib.pyplot as plt

# Image loading
img = io.imread('tst1.bmp')
# Grayscale conversion
img = color.rgb2gray(img)
# add Gaussian noise to the image
# the variance sets the width of the gaussian
img = util.random_noise(img, mode='gaussian', mean=0, var = (10/255)**2)
img = img*255
# Displaying the grayscale image
plt.figure(),plt.imshow(img,cmap='gray', vmin = 0, vmax = 255), plt.colorbar()
hist,_ = np.histogram(img,range(0,256))
plt.figure(figsize=(10,10)),plt.plot(hist)

segm = np.zeros(img.shape)
segm[img<25] = 0
segm[np.logical_and(img>=25, img<75)] = 1
segm[np.logical_and(img>=75, img<125)] = 2
segm[np.logical_and(img>=125, img<175)] = 3
segm[img>175] = 4
plt.figure(figsize=(10,10)),plt.imshow(segm,cmap='gray', vmin = 0, vmax = 4), plt.colorbar()

And this is my result:
filter
It looks like I still have some salt&pepper noise present, how can I eliminate it? I don’t want to use any Skimage functions built-in Python, I need to use code made by me.

#2

Just a note:
Without prior knowledge it is impossible to remove stochastic noise, you can at best reduce it with a certain impact on the image structures.

#3

Yeah, at least to reduce it, I have this code but it’s not OK because it did not do the segmentation on the new filtered image.

from skimage.morphology import disk
from skimage.filters import rank
# Define the neighbourd disk whose mean you would like to take for mean filtering
selem = disk(4)
# Convert image to uint8 for converting the image into required format
img = util.img_as_ubyte(img/255)
# Get the result by performing mean filter
normal_result = rank.mean(img, selem=selem)


fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(10, 10),
                         sharex=True, sharey=True)
ax = axes.ravel()


titles = ['Original','After Mean filter']
imgs = [img, normal_result]
for n in range(0, len(imgs)):
    ax[n].imshow(imgs[n], cmap=plt.cm.gray)
    ax[n].set_title(titles[n])
    ax[n].axis('off')


plt.tight_layout()
plt.show()