Basic image manipulation with arrays in NumPy - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

What is the output shape after flipping the image array?

Given a 3D NumPy array representing an RGB image with shape (100, 200, 3), what will be the shape of the array after applying np.flipud(image)?

NumPy

import numpy as np
image = np.zeros((100, 200, 3))
flipped = np.flipud(image)
print(flipped.shape)

A (100, 3, 200)

B (100, 200, 3)

C (3, 100, 200)

D (200, 100, 3)

💡 Hint

Flipping up-down reverses rows but does not change the shape.

The np.flipud function flips the array vertically (up-down) but keeps the shape the same. So the output shape remains (100, 200, 3).

❓ data_output

intermediate

What is the pixel value after increasing brightness?

Given a grayscale image array with pixel value 120, what will be the new pixel value after adding 50 and clipping to 255?

NumPy

import numpy as np
pixel = np.array([120])
bright_pixel = np.clip(pixel + 50, 0, 255)
print(bright_pixel[0])

A 170

B 255

C 120

D 50

💡 Hint

Adding 50 to 120 gives 170, which is less than 255, so clipping does not change it.

120 + 50 = 170, which is within the allowed range 0-255, so the pixel value becomes 170.

❓ visualization

advanced

Which option shows the correct code to convert an RGB image to grayscale?

You have an RGB image array with shape (height, width, 3). Which code correctly converts it to a grayscale image using the luminosity method?

NumPy

import numpy as np

# image is a numpy array with shape (height, width, 3)

A grayscale = np.dot(image[..., :3], [0.2989, 0.5870, 0.1140])

B grayscale = np.mean(image, axis=2)

C grayscale = image[:, :, 0] * 0.3 + image[:, :, 1] * 0.59 + image[:, :, 2] * 0.11

D grayscale = np.sum(image, axis=2) / 3

💡 Hint

The luminosity method uses weighted sums of RGB channels.

Option A uses the correct weights for red, green, and blue channels to compute grayscale using the luminosity method.

🔧 Debug

advanced

What error does this code raise when trying to invert a binary image?

Consider this code snippet to invert a binary image (0s and 1s):

inverted = 1 - image

If image is a NumPy array of dtype bool, what error occurs?

NumPy

import numpy as np
image = np.array([[True, False], [False, True]], dtype=bool)
inverted = 1 - image

A TypeError: unsupported operand type(s) for -: 'bool' and 'bool'

B ValueError: operands could not be broadcast together

C No error, output is a boolean array inverted correctly

D TypeError: unsupported operand type(s) for -: 'int' and 'bool'

💡 Hint

Subtracting a boolean array from an integer can cause type errors.

Subtracting a boolean array from an integer raises a TypeError because NumPy does not support subtraction between int and bool directly.

🚀 Application

expert

How many pixels change after thresholding an image?

Given a grayscale image array of shape (50, 50) with pixel values from 0 to 255, you apply thresholding:

thresholded = (image > 128).astype(int)

If exactly 1200 pixels have values greater than 128, how many pixels in thresholded are 1?

NumPy

import numpy as np
np.random.seed(0)
image = np.random.randint(0, 256, (50, 50))
thresholded = (image > 128).astype(int)
count_ones = np.sum(thresholded)
print(count_ones)

A 2500

B 1300

C 1200

D 128

💡 Hint

Thresholding sets pixels > 128 to 1, so count of 1s equals count of pixels > 128.

The thresholded array has 1s exactly where the original image pixels are greater than 128, so the count is 1200.