MCA III Semester
Digital Image Processing Lab Assignments
1. Write a program to demonstrate quantization.
import matplotlib.pyplot as plt
import numpy as np
im = plt.imread('cameraman.tif')
plt.imshow(im)
plt.title('Original Image')
plt.show()
im_d = im.astype(float)
c0 = np.mod(im_d, 2)
plt.imshow(c0)
plt.title('Zero bit Image')
plt.show()
c1 = np.mod(np.floor(im_d/2), 2)
plt.imshow(c1)
plt.title('First bit Image')
plt.show()
c2 = np.mod(np.floor(im_d/4), 2)
plt.imshow(c2)
plt.title('Second bit Image')
plt.show()
c3 = np.mod(np.floor(im_d/8), 2)
plt.imshow(c3)
plt.title('Third bit Image')
plt.show()
c4 = np.mod(np.floor(im_d/16), 2)
plt.imshow(c4)
plt.title('Fourth bit Image')
plt.show()
c5 = np.mod(np.floor(im_d/32), 2)
plt.imshow(c5)
plt.title('Fifth bit Image')
plt.show()
c6 = np.mod(np.floor(im_d/64), 2)
plt.imshow(c6)
plt.title('Sixth bit Image')
plt.show()
c7 = np.mod(np.floor(im_d/128), 2)
plt.imshow(c7)
plt.title('Seventh bit Image')
plt.show()
cc = 2*(2*(2*(2*(2*(2*(2*c7+c6)+c5)+c4)+c3)+c2)+c1)+c0
plt.imshow(cc.astype(np.uint8))
plt.title('Reconstructed Original Image')
plt.show()
2. Write a program to demonstrate down sampling.
import cv2
import matplotlib.pyplot as plt
im = cv2.imread('pout.tif', 0)
plt.imshow(im, cmap='gray')
plt.title('Original image')
plt.show()
plt.hist(im.ravel(), bins=256, range=[0,256])
plt.axis('tight')
plt.title('Image Histogram')
plt.show()
im_h = cv2.equalizeHist(im)
plt.imshow(im_h, cmap='gray')
plt.title('Histogram equalized image')
plt.show()
plt.hist(im_h.ravel(), bins=256, range=[0,256])
plt.axis('tight')
plt.title('Histogram of equalized image')
plt.show()
3. Write a program to perform bit plain slicing.
import cv2
import numpy as np
import matplotlib.pyplot as plt
I = cv2.imread('cameraman.tif', cv2.IMREAD_GRAYSCALE)
plt.subplot(2,2,1)
plt.imshow(I, cmap='gray')
plt.title('Original Image')
H = np.zeros((20, 20))
H[10, :] = 1
H = H / np.sum(H)
MotionBlur = cv2.filter2D(I, -1, H,
borderType=cv2.BORDER_REPLICATE)
plt.subplot(2,2,2)
plt.imshow(MotionBlur, cmap='gray')
plt.title('Motion Blurred Image')
H = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (10, 10))
blurred = cv2.filter2D(I, -1, H,
borderType=cv2.BORDER_REPLICATE)
plt.subplot(2,2,3)
plt.imshow(blurred, cmap='gray')
plt.title('Blurred Image')
H = np.array([[0, -1, 0], [-1, 5, -1], [0, -1, 0]])
sharpened = cv2.filter2D(I, -1, H,
borderType=cv2.BORDER_REPLICATE)
plt.subplot(2,2,4)
plt.imshow(sharpened, cmap='gray')
plt.title('Sharpened Image')
plt.show()
4. Write a program to demonstrate histogram equalization.
import numpy as np
import matplotlib.pyplot as plt
from scipy.fftpack import fftshift, fft2
from PIL import Image
a = np.zeros((256, 256))
a[78:178, 78:178] = 1
plt.imshow(a), plt.title('Image with a square')
plt.show()
af = fftshift(fft2(a))
plt.imshow(np.log(1 + np.abs(af))), plt.title('Fourier transform of square
image')
plt.show()
x, y = np.meshgrid(np.arange(1, 257), np.arange(1, 257))
b = (x + y < 329) & (x + y > 182) & (x - y > -67) & (x - y < 73)
plt.imshow(b), plt.title('Rotated square image')
plt.show()
bf = fftshift(fft2(b))
plt.imshow(np.log(1 + np.abs(bf))), plt.title('Fourier transform of
rotated square image')
plt.show()
x, y = np.meshgrid(np.arange(-128, 218), np.arange(-128, 128))
c = np.array(Image.open('cameraman.tif'))
plt.imshow(c), plt.title('Cameraman image')
plt.show()
cf = fftshift(fft2(c))
plt.imshow(np.log(1 + np.abs(cf))), plt.title('Fourier transform of
Cameraman image')
plt.show()
p, q = np.meshgrid(np.arange(-128, 218), np.arange(-128, 128))
r = np.sqrt(p**2 + q**2)
s = (r < 15)
cf = fftshift(fft2(s))
plt.imshow(s), plt.title('Circle')
plt.show()
plt.imshow(np.log(1 + np.abs(cf))), plt.title('Fourier transform of
Circle')
plt.show()
5. Write a program to smoothing and sharpening using spatial filters.
import numpy as np
import matplotlib.pyplot as plt
from scipy.fftpack import fftshift, fft2, ifft2
cm = plt.imread('cameraman.tif')
plt.imshow(cm)
plt.title('Input image')
plt.show()
cf = fftshift(fft2(cm))
plt.imshow(np.log(1 + np.abs(cf)))
plt.title('Fourier transform of input image')
plt.show()
x, y = np.meshgrid(np.arange(-128, 128), np.arange(-128, 128))
z = np.sqrt(x**2 + y**2)
c = (z < 15) # Ideal lowpass filter
cfl = cf * c
plt.imshow(np.log(1 + np.abs(cfl)))
plt.title('Ideal lowpass filter')
plt.show()
cfli = ifft2(cfl)
plt.imshow(np.abs(cfli))
plt.title('Image smoothing using ideal lowpass filter')
plt.show()
c = (z > 15) # Ideal highpass filter
cfh = cf * c
plt.imshow(np.log(1 + np.abs(cfh)))
plt.title('Ideal highpass filter')
plt.show()
cfhi = ifft2(cfh)
plt.imshow(np.abs(cfhi))
plt.title('Image sharpening using ideal highpass filter')
plt.show()
6. Write a program to generate Fourier transform of an image.
# Assignment 6: Noise generation
import cv2
import numpy as np
# Load the image
tw = cv2.imread('cameraman.tif')
cv2.imshow('Input image', tw)
cv2.waitKey(0)
cv2.destroyAllWindows()
# Add salt and pepper noise
t_sp = cv2.imread('cameraman.tif')
noise = np.random.randint(0, 2, size=t_sp.shape[:2])
t_sp[noise == 0] = 0
t_sp[noise == 1] = 255
cv2.imshow('Image with salt & pepper noise', t_sp)
cv2.waitKey(0)
cv2.destroyAllWindows()
# Add Gaussian noise
t_ga = cv2.imread('cameraman.tif')
mean = 0
stddev = 50
noise = np.random.normal(mean, stddev,
size=t_ga.shape).astype(np.uint8)
t_ga = cv2.add(t_ga, noise)
cv2.imshow('Image with Gaussian noise', t_ga)
cv2.waitKey(0)
cv2.destroyAllWindows()
# Add speckle noise
t_spk = cv2.imread('cameraman.tif')
noise = np.random.randn(*t_spk.shape).astype(np.uint8)
t_spk = t_spk + t_spk * noise
cv2.imshow('Image with speckle noise', t_spk)
cv2.waitKey(0)
cv2.destroyAllWindows()
# Add periodic noise
t_pn = cv2.imread('cameraman.tif')
s = t_pn.shape
x, y = np.meshgrid(np.arange(s[1]), np.arange(s[0]))
p = np.sin(x / 3 + y / 5) + 1
t_pn = (t_pn.astype(np.float32) + p / 2) / 2
cv2.imshow('Image with periodic noise', t_pn)
cv2.waitKey(0)
cv2.destroyAllWindows()
7. Write a program to perform image smoothing and sharpening
using Ideal filters.
# Assignment 7: Noise removal
import cv2
import numpy as np
# Read the input image
t = cv2.imread('cameraman.tif', 0)
cv2.imshow('Input image', t)
cv2.waitKey(0)
cv2.destroyAllWindows()
# Add salt and pepper noise to the image
t_sp = cv2.imread('cameraman.tif', 0)
t_sp = cv2.cvtColor(t_sp, cv2.COLOR_BGR2GRAY)
noise = np.random.randint(0, 2, size=t_sp.shape)
t_sp[noise == 0] = 0
t_sp[noise == 1] = 255
cv2.imshow('Image with salt & pepper noise', t_sp)
cv2.waitKey(0)
cv2.destroyAllWindows()
# Apply 3x3 averaging filter
a3 = np.ones((3, 3), dtype=np.float32) / 9
t_sp_a3 = cv2.filter2D(t_sp, -1, a3)
cv2.imshow('3x3 averaging', t_sp_a3)
cv2.waitKey(0)
cv2.destroyAllWindows()
# Apply 7x7 averaging filter
a7 = np.ones((7, 7), dtype=np.float32) / 49
t_sp_a7 = cv2.filter2D(t_sp, -1, a7)
cv2.imshow('7x7 averaging', t_sp_a7)
cv2.waitKey(0)
cv2.destroyAllWindows()
# Apply median filtering
t_sp_m3 = cv2.medianBlur(t_sp, 3)
cv2.imshow('Median filtering', t_sp_m3)
cv2.waitKey(0)
cv2.destroyAllWindows()
8. Write a program to add different noise to an image.
# Assignment 8: Dilation and erosion
import cv2
import numpy as np
import matplotlib.pyplot as plt
# Read the image
t = cv2.imread('text.jpg', 0)
# Convert the image to binary
t = cv2.threshold(t, 127, 255, cv2.THRESH_BINARY)[1]
# Create a square structuring element
sq = np.ones((2, 2), np.uint8)
# Perform dilation
td = cv2.dilate(t, sq, iterations=1)
# Display the original image
plt.subplot(1, 4, 1)
plt.imshow(t, cmap='gray')
plt.title('Original image')
# Display the dilated image
plt.subplot(1, 4, 2)
plt.imshow(td, cmap='gray')
plt.title('Dilated image')
# Perform erosion
ce = cv2.erode(t, sq, iterations=1)
# Display the eroded image
plt.subplot(1, 4, 3)
plt.imshow(ce, cmap='gray')
plt.title('Eroded image')
# Create a larger square structuring element
sq = np.ones((3, 3), np.uint8)
# Perform erosion with the larger structuring element
ce = cv2.erode(t, sq, iterations=1)
# Display the eroded image with the larger structuring element
plt.subplot(1, 4, 4)
plt.imshow(ce, cmap='gray')
plt.title('Eroded image')
# Show all the plots
plt.show()
9. Write a program to remove noise from an image.
# Assignment 9: Boundary detection
import cv2
import numpy as np
import matplotlib.pyplot as plt
im = cv2.imread('rice.png', 0)
r = cv2.threshold(im, 127, 255, cv2.THRESH_BINARY)[1]
sq = np.ones((3,3), np.uint8)
re = cv2.erode(r, sq)
r_int = cv2.bitwise_and(r, cv2.bitwise_not(re))
plt.subplot(2,2,1),plt.imshow(im, cmap='gray'), plt.title('Input image')
plt.subplot(2,2,2),plt.imshow(r_int, cmap='gray'), plt.title('Internal
boundary')
rd = cv2.dilate(r, sq)
r_ext = cv2.bitwise_and(rd, cv2.bitwise_not(r))
r_grad = cv2.bitwise_and(rd, cv2.bitwise_not(re))
plt.subplot(2,2,3),plt.imshow(r_ext, cmap='gray'), plt.title('External
boundary')
plt.subplot(2,2,4),plt.imshow(r_grad, cmap='gray'),
plt.title('Morphological gradient')
plt.show()
10. Write a program to demonstrate dilation and erosion.
# Assignment 10: Morphological Opening and closing
import cv2
import numpy as np
# Read the image
c = cv2.imread('circles.jpg', 0)
# Convert the image to binary
c = cv2.threshold(c, 127, 255, cv2.THRESH_BINARY)[1]
# Generate random noise
x = np.random.rand(*c.shape)
# Apply noise to the image
d1 = np.where(x <= 0.05)
d2 = np.where(x >= 0.95)
c[d1] = 0
c[d2] = 1
# Display the image with noise
cv2.imshow('Circle image with noise', c)
cv2.waitKey(0)
# Define the structuring elements
sq = np.ones((3, 3), np.uint8)
cr = np.array([[0, 1, 0], [1, 1, 1], [0, 1, 0]], np.uint8)
# Perform closing and opening operations with square structuring
element
cf1 = cv2.morphologyEx(c, cv2.MORPH_CLOSE, sq)
cf1 = cv2.morphologyEx(cf1, cv2.MORPH_OPEN, sq)
# Display the result
cv2.imshow('Closing & Opening with Square SE', cf1)
cv2.waitKey(0)
# Perform closing and opening operations with cross structuring
element
cf2 = cv2.morphologyEx(c, cv2.MORPH_CLOSE, cr)
cf2 = cv2.morphologyEx(cf2, cv2.MORPH_OPEN, cr)
# Display the result
cv2.imshow('Closing & Opening with cross SE', cf2)
cv2.waitKey(0)
# Close all windows
cv2.destroyAllWindows()
