Blurring and smoothing (Gaussian, median, bilateral) in Computer Vision - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of Gaussian Blur on a simple image array

Given the following 3x3 grayscale image array and applying a Gaussian blur with kernel size 3 and sigma 1, what is the center pixel value after blurring?

Computer Vision

import cv2
import numpy as np

image = np.array([[10, 20, 30],
                  [20, 30, 40],
                  [30, 40, 50]], dtype=np.uint8)

blurred = cv2.GaussianBlur(image, (3, 3), 1)
center_pixel = blurred[1, 1]
print(center_pixel)

A 30

B 28

C 29

D 31

💡 Hint

Gaussian blur smooths pixels weighted by a Gaussian kernel; center pixel is a weighted average.

The Gaussian blur with kernel size 3 and sigma 1 computes a weighted average around the center pixel. The exact value is 29.

🧠 Conceptual

intermediate

Effect of median blur on salt-and-pepper noise

Which statement best describes the effect of applying a median blur filter on an image corrupted with salt-and-pepper noise?

A Median blur replaces each pixel with the average of its neighbors, smoothing the noise.

B Median blur replaces each pixel with the median of its neighbors, effectively removing salt-and-pepper noise.

C Median blur sharpens edges while reducing noise.

D Median blur applies a Gaussian weighted average, which is ineffective against salt-and-pepper noise.

💡 Hint

Think about how median values handle extreme pixel values.

Median blur replaces each pixel with the median of its neighbors, which removes isolated extreme values like salt-and-pepper noise effectively.

❓ Model Choice

advanced

Choosing the best smoothing filter for edge preservation

You want to smooth an image to reduce noise but keep edges sharp and clear. Which filter is the best choice?

A Bilateral filter

B Median blur

C Gaussian blur

D Average blur

💡 Hint

Consider which filter uses both spatial and intensity information.

Bilateral filter smooths while preserving edges by considering both pixel distance and intensity differences.

❓ Hyperparameter

advanced

Effect of increasing sigma in Gaussian blur

What happens to the image when you increase the sigma value in a Gaussian blur filter while keeping the kernel size constant?

A The filter ignores pixels farther from the center.

B The image noise increases due to less smoothing.

C The image becomes sharper with more details preserved.

D The image becomes more blurred with stronger smoothing effect.

💡 Hint

Sigma controls the spread of the Gaussian kernel.

Increasing sigma makes the Gaussian kernel wider, causing more smoothing and blur.

🔧 Debug

expert

Identifying error in bilateral filter usage

What error will this code produce when applying bilateral filter with the given parameters? import cv2 import numpy as np image = np.ones((5,5), dtype=np.uint8)*100 blurred = cv2.bilateralFilter(image, d=0, sigmaColor=75, sigmaSpace=75) print(blurred)

Computer Vision

import cv2
import numpy as np

image = np.ones((5,5), dtype=np.uint8)*100
blurred = cv2.bilateralFilter(image, d=0, sigmaColor=75, sigmaSpace=75)
print(blurred)

A No error, prints the blurred image array

B cv2.error: d must be positive or -1

C TypeError: unsupported operand type(s) for *: 'int' and 'str'

D ValueError: sigmaColor must be less than 50

💡 Hint

Check the meaning of parameter d in bilateralFilter.

No error occurs because when d is non-positive (including 0), it is automatically computed from sigmaSpace.