05/05/25, 15.19 1-Soshum-M7-Image-Processing.

ipynb - Colab

keyboard_arrow_down Introduction to Image Processing in Python

Praktikum ini memberikan pengenalan teknik pemrosesan gambar dasar menggunakan library computer vision OpenCV dan
beberapa library analisis data standar dengan Python. Pengetahuan tentang pemrosesan gambar dan pemrograman Python
tidak diperlukan untuk workshop ini, tetapi akan membantu Anda untuk memahami apa yang Anda kerjakan.

The source of this notebook is located at https://github.com/xn2333/OpenCV/blob/master/Seminar_Image_Processing_in_Python.ipynb

Useful intro about Colab

Useful intro about OpenCV

keyboard_arrow_down Step1: Load the Dependencies

Anda akan memanggil beberapa libraries yang akan digunakan selama praktikum ini: numpy, pandas, cv2, skimage, PIL,
matplotlib

Numpy is an array manipulation library, used for linear algebra, Fourier transform, and random number capabilities.
Pandas is a library for data manipulation and data analysis.
CV2 is a library for computer vision tasks.
Skimage is a library which supports image processing applications on python.
Matplotlib is a library which generates figures and provides graphical user interface toolkit.

import numpy as np
import pandas as pd
import cv2 as cv
from google.colab.patches import cv2_imshow # for image display
from skimage import io
from PIL import Image
import matplotlib.pylab as plt

keyboard_arrow_down Step2: Read Image from Urls

Pada langkah ini kita akan membaca gambar dari url, dan menampilkannya menggunakan openCV, perhatikan perbedaannya
saat membaca gambar dalam format RGB dan BGR. Channel warna input default pada library openCV adalah dalam format
BGR.

RGB? Red Green Blue. RGB adalah channel warna standar yang digunakan oleh sebuah layar/gambar berwarna.

# Create a list to store the urls of the images

urls = ["https://iiif.lib.ncsu.edu/iiif/0052574/full/800,/0/default.jpg",
"https://iiif.lib.ncsu.edu/iiif/0016007/full/800,/0/default.jpg"]
# Read and display the image
# loop over the image URLs, you could store several image urls in the list

for url in urls:

image = io.imread(url)
image_2 = cv.cvtColor(image, cv.COLOR_BGR2RGB)
final_frame = cv.hconcat((image, image_2))
cv2_imshow(final_frame)
print('\n')

https://colab.research.google.com/drive/1LhCBqGP-0lXVkZuAOn80tMCcf6kFFa0L#scrollTo=XUMXfKr5gMl9&printMode=true 1/9
05/05/25, 15.19 1-Soshum-M7-Image-Processing.ipynb - Colab

keyboard_arrow_down TODO #1: Read an image from a URL and display it

Image source examples:

Place Kitten - use the base Place Kitten URL followed by a width and height separated by backslashes ''/''. For example, use the URL
https://placekitten.com/500/300 to fetch a cat image with a width of 500px and height of 300px.

NC State University Libraries Special Collections - browse the site to find an image thumbnail. Right-click on the thumbnail and select "Copy
Image Address". The address will look like this: https://iiif.lib.ncsu.edu/iiif/0051230/square/300,/0/default.jpg . Replace the word
"square" with the word "full" and replace "300" with "800" to access the full image at a width of 800px.

Google Image search - search for an image. Left-click one of the returned images, then right-click on the full image, and then select "Copy
Image Address".

## Silahkan lakukan langkah seperti diatas, dengan gambar Anda sendiri dengan menghapus tanda #
## pada kode dibawah ini dan mengisi url dari gambar yang akan Anda gunakan.

# url =
# myImg = io.imread(url)
# cv2_imshow(cv.cvtColor(myImg, cv.COLOR_BGR2RGB))

keyboard_arrow_down Step3: Image Contours and Histograms

# Check the image matrix data type (could know the bit depth of the image)
print(image.dtype)
# Check the height of image
print(image.shape[0])
# Check the width of image

https://colab.research.google.com/drive/1LhCBqGP-0lXVkZuAOn80tMCcf6kFFa0L#scrollTo=XUMXfKr5gMl9&printMode=true 2/9
05/05/25, 15.19 1-Soshum-M7-Image-Processing.ipynb - Colab
print(image.shape[1])
# Check the number of channels of the image
print(image.shape[2])

uint8
571
800
3

keyboard_arrow_down Generate Histogram of color image and grayscale image

Terkadang Anda ingin meningkatkan kontras pada gambar atau memperluas kontras di wilayah tertentu sambil mengorbankan detail dalam
warna yang tidak terlalu bervariasi, atau tidak penting. Alat yang baik untuk menemukan wilayah yang menarik adalah dengan menggunakan
histogram gambar. Untuk membuat histogram dari data gambar kita, kita menggunakan fungsi matplot.pylab hist ().

Info selengkapnya: Histogram

Display the histogram of all the pixels in the color image

plt.hist(image.ravel(),bins = 256, range = [0,256])

plt.show()

Menampilkan histogram dari channel R, G, B. Kita dapat mengamati bahwa saluran hijau memiliki banyak pixel pada nilai 255 dimana nilai
255 ini mewakili warna putih pada gambar

color = ('b','g','r')
for i,col in enumerate(color):
histr = cv.calcHist([image],[i],None,[256],[0,256])
plt.plot(histr,color = col)
plt.xlim([0,256])
plt.show()

 

https://colab.research.google.com/drive/1LhCBqGP-0lXVkZuAOn80tMCcf6kFFa0L#scrollTo=XUMXfKr5gMl9&printMode=true 3/9
05/05/25, 15.19 1-Soshum-M7-Image-Processing.ipynb - Colab

gray_image = cv.cvtColor(image, cv.COLOR_BGR2GRAY)

cv2_imshow(gray_image)

 

# Plot the histogram of the gray image. We could observe that the frequency of
# the image hist has decreased ~ 1/3 of the histogram of color image
plt.hist(gray_image.ravel(),bins = 256, range = [0, 256])
plt.show()

keyboard_arrow_down TODO # 2: DISPLAY THE GRAYSCALE OF YOUR COLOR IMAGE AND GENERATE HISTOGRAM

## Silahkan lakukan langkah seperti diatas, dengan gambar Anda sendiri dengan menghapus tanda #
## pada kode dibawah ini

#myGrayImg = cv.cvtColor(myImg, cv.COLOR_BGR2GRAY)

#cv2_imshow()

keyboard_arrow_down Find image contour of the grayscale image

Method 1: Use the matplotlib. contour

More Info: matplotlib contour

plt.contour(gray image, origin = "image")

https://colab.research.google.com/drive/1LhCBqGP-0lXVkZuAOn80tMCcf6kFFa0L#scrollTo=XUMXfKr5gMl9&printMode=true 4/9
05/05/25, 15.19 1-Soshum-M7-Image-Processing.ipynb - Colab
plt.contour(gray_image, origin image )

<matplotlib.contour.QuadContourSet at 0x7b5e7efc0d90>

Method 2: Use the openCV lib

More info: Contour

# Set threshold for the countour detection

ret, thresh = cv.threshold(gray_image,150,255,0)
contours, hierarchy = cv.findContours(thresh, cv.RETR_TREE, cv.CHAIN_APPROX_SIMPLE)
cv.drawContours(image, contours, -1, (0, 255, 0), 3)
plt.imshow(image)

<matplotlib.image.AxesImage at 0x7b5e7ee94750>

 

keyboard_arrow_down TODO # 3: Find the contour of your own image

## Silahkan lakukan langkah seperti diatas, dengan gambar Anda sendiri dengan menghapus tanda #
## pada kode dibawah ini

# plt.contour(myGrayImg)

keyboard_arrow_down Step4: Grayscale Transform and Histogram Equalization

keyboard_arrow_down Grayscale Transformation

Bagian ini memberikan beberapa contoh melakukan transformasi matematis pada gambar grayscale

# This is an inverse operation of the grayscale image, you could see that the
# bright pixels become dark, and the dark pixels become bright
im2 = 255 - gray_image
cv2_imshow(im2)

https://colab.research.google.com/drive/1LhCBqGP-0lXVkZuAOn80tMCcf6kFFa0L#scrollTo=XUMXfKr5gMl9&printMode=true 5/9
05/05/25, 15.19 1-Soshum-M7-Image-Processing.ipynb - Colab

 

# Another transform of the image, after adding a constant,

# all the pixels become brighter and a hazing-like effect of the image is generated
im3 = (100.0/255)*gray_image + 100
cv2_imshow(im3)

 

# The lightness level of the gray_image decreases after this step

im4 = 255.0*(gray_image/255.0)**2
cv2_imshow(im4)

https://colab.research.google.com/drive/1LhCBqGP-0lXVkZuAOn80tMCcf6kFFa0L#scrollTo=XUMXfKr5gMl9&printMode=true 6/9
05/05/25, 15.19 1-Soshum-M7-Image-Processing.ipynb - Colab

 

keyboard_arrow_down Histogram Equalization

This section demonstrates histogram equalization on a dark image. This transform flattens the gray-level histogram so that all intensities are
as equally common as possible. The transform function is a cumulative distribution function (cdf) of the pixel values in the image
(normalized to map the range of pixel values to the desired range). This example uses image 4 (im4).

# function of histogram equalization

def histeq(im, nbr_bins = 256):
""" Histogram equalization of a grayscale image. """
# get the image histogram
imhist, bins = np.histogram(im.flatten(), nbr_bins, [0, 256])
cdf = imhist.cumsum() # cumulative distribution function
cdf = imhist.max()*cdf/cdf.max() #normalize
cdf_mask = np.ma.masked_equal(cdf, 0)
cdf_mask = (cdf_mask - cdf_mask.min())*255/(cdf_mask.max()-cdf_mask.min())
cdf = np.ma.filled(cdf_mask,0).astype('uint8')
return cdf[im.astype('uint8')]

# apply the function on your dark image to increase the contrast

# we could observe that the contrast of the black background has increased
im5 = histeq(im4)
cv2_imshow(im5)

https://colab.research.google.com/drive/1LhCBqGP-0lXVkZuAOn80tMCcf6kFFa0L#scrollTo=XUMXfKr5gMl9&printMode=true 7/9
05/05/25, 15.19 1-Soshum-M7-Image-Processing.ipynb - Colab

 

# Extra: try to visualize the histogram of the image after histogram equalization
# Before histogram equalization
plt.hist(im4.ravel(),bins = 256, range = [0, 256])
plt.show()

 

# After histogram equalization

plt.hist(im5.ravel(),bins = 256, range = [0, 256])
plt.show()

https://colab.research.google.com/drive/1LhCBqGP-0lXVkZuAOn80tMCcf6kFFa0L#scrollTo=XUMXfKr5gMl9&printMode=true 8/9
05/05/25, 15.19 1-Soshum-M7-Image-Processing.ipynb - Colab

keyboard_arrow_down Step5: Finding Edges by Highpass Filtering in FFT

Bagian ini mendemonstrasikan melakukan high pass filter untuk menghilangkan komponen frekuensi rendah, menghasilkan gambar yang
tajam yang berisi tepinya

# Blur the grayscale image by a Guassian filter with kernel size of 10

imBlur = cv.blur(gray_image,(5,5))
# Transform the image to frequency domain
f = np.fft.fft2(imBlur)
# Bring the zero-frequency component to the center
fshift = np.fft.fftshift(f)
magnitude_spectrum = 30*np.log(np.abs(fshift))

plt.subplot(121),plt.imshow(imBlur, cmap = 'gray')

 
plt.title('Input Image'), plt.xticks([]), plt.yticks([])
plt.subplot(122),plt.imshow(magnitude_spectrum, cmap = 'gray')
plt.title('Magnitude Spectrum'), plt.xticks([]), plt.yticks([])
plt.show()

https://colab.research.google.com/drive/1LhCBqGP-0lXVkZuAOn80tMCcf6kFFa0L#scrollTo=XUMXfKr5gMl9&printMode=true 9/9