how to remove salt and pepper noise from images using python?

Question

I have tried to implement the following algorithm but the resulting image looks the same.

Step 1: Read Noisy Image.

Step 2: Select 2D window of size 3x3 with centre element as processing pixel. Assume that the pixel being processed is P ij .

Step 3: If P ij is an uncorrupted pixel (that is, 0< P ij <255), then its value is left unchanged.

Step 4: If P ij = 0 or P ij = 255, then P ij is a corrupted pixel.

Step 5: If 3/4 th or more pixels in selected window are noisy then increase window size to 5x5. Step 6: If all the elements in the selected window are 0‟s and 255‟s, then replace P ij with the mean of the elements in the window else go to step 7.

Step 7: Eliminate 0‟s and 255‟s from the selected window and find the median value of the remaining elements. Replace Pij with the median value.

Step 8: Repeat steps 2 to 6 until all the pixels in the entire image are processed.

Here is my code. Please suggest improvements.

import Image

im=Image.open("no.jpg")
im = im.convert('L')

for i in range(2,im.size[0]-2):
    for j in range(2,im.size[1]-2):
        b=[]
        if im.getpixel((i,j))>0 and im.getpixel((i,j))<255:
            pass
        elif im.getpixel((i,j))==0 or im.getpixel((i,j))==255:
            c=0
            for p in range(i-1,i+2):
                for q in range(j-1,j+2):
                    if im.getpixel((p,q))==0 or im.getpixel((p,q))==255: 
                        c=c+1
            if c>6:
                c=0
                for p in range(i-2,i+3):
                    for q in range(j-2,j+3):
                        b.append(im.getpixel((p,q)))
                        if im.getpixel((p,q))==0 or im.getpixel((p,q))==255:
                            c=c+1
                if c==25:
                    a=sum(b)/25
                    print a
                    im.putpixel((i,j),a)
                else:
                    p=[]
                    for t in b:
                        if t not in (0,255):
                            p.append(t)
                    p.sort()
                    im.putpixel((i,j),p[len(p)/2])
            else:
                b1=[]
                for p in range(i-1,i+2):
                    for q in range(j-1,j+2):
                        b1.append(im.getpixel((p,q)))
                im.putpixel((i,j),sum(b1)/9)

im.save("nonoise.jpg")

How did you determine that the two images were the same. Visual inspection or an image difference? They might look similar except for a few pixels. One concern with looking at your code is that you append to your temp arrays (b and b1) regardless of whether the pixel was 0, or 255. If you want the median of only the valid pixels you need to have an additional test before you append. Secondly, you are computing something like a mean, instead of median (numpy.median) — Paul, Mar 10 '14 at 12:18

score 7 · Answer 1 · answered Mar 08 '14 at 17:36

7

You should use median filter, it is easy to implement and work very fine for salt and pepper noise.

answered Mar 08 '14 at 17:36

Olivier A

842
4
8

score 3 · Answer 2 · answered Sep 30 '16 at 05:34

As Olivier suggested, the median filter provides the best result.

Here is the code I generated for adding salt and pepper noise into an image. The code is for python with OpenCV 3.0.0 :

import numpy as np
import cv2

img = cv2.imread('3.jpg', 1)
row,col,ch = img.shape
p = 0.5
a = 0.009
noisy = img

  # Salt mode
num_salt = np.ceil(a * img.size * p)
coords = [np.random.randint(0, i - 1, int(num_salt))
          for i in img.shape]
noisy[coords] = 1

  # Pepper mode
num_pepper = np.ceil(a * img.size * (1. - p))
coords = [np.random.randint(0, i - 1, int(num_pepper))
          for i in img.shape]
noisy[coords] = 0

cv2.imshow('noisy', noisy)

Here is the code to use the median filter:

median_blur= cv2.medianBlur(noisy, 3)
cv2.imshow('median_blur', median_blur)  

cv2.waitKey()
cv2.destroyAllWindows()

The window used to blur the noisy image can be modified as per requirement.

score 0 · Answer 3 · answered Mar 08 '14 at 14:44

0

What does your input image look like? You algorithm assumes only pixel value 0 and 255 is noise. If your noisy pixels actually have values other than that, your algorithm will not do anything and you might see output looks identical to input.

answered Mar 08 '14 at 14:44

ycyeh

381
2
4

Even though I have some pixels with values 0 and 255, the output image is the same as the input image. – user3395151 Mar 10 '14 at 08:33

how to remove salt and pepper noise from images using python?

3 Answers3