Image filtering (gaussian_filter) in SciPy - Step-by-Step Execution

Concept Flow - Image filtering (gaussian_filter)

Input Image Array

↓

Apply gaussian_filter

↓

Smooth Image Output

↓

Use or Display Result

Start with an image array, apply gaussian_filter to smooth it, then get the filtered image for use or display.

Execution Sample

SciPy

import numpy as np
from scipy.ndimage import gaussian_filter

image = np.array([[10, 10, 10],
                  [10, 50, 10],
                  [10, 10, 10]])
filtered = gaussian_filter(image, sigma=1)

This code smooths a small 3x3 image using a Gaussian filter with sigma=1.

Execution Table

Step	Action	Input Value	Filter Parameter	Output Value
1	Input image created	[[10,10,10],[10,50,10],[10,10,10]]	sigma=1	Same as input
2	Apply gaussian_filter	Image array	sigma=1	Smoothed image array
3	Calculate weighted average for center pixel	Center=50 neighbors=10	sigma=1	Center pixel approx 22.5
4	Calculate weighted average for corner pixel	Corner=10 neighbors=10 and 50	sigma=1	Corner pixel approx 12.5
5	Filtered image ready	N/A	N/A	[[12.5, 15.6, 12.5],[15.6, 22.5, 15.6],[12.5, 15.6, 12.5]]

💡 Filtering complete, output is smoothed image array

Variable Tracker

Variable	Start	After gaussian_filter	Final
image	[[10,10,10],[10,50,10],[10,10,10]]	[[10,10,10],[10,50,10],[10,10,10]]	Same as start
filtered	None	N/A	[[12.5, 15.6, 12.5],[15.6, 22.5, 15.6],[12.5, 15.6, 12.5]]

Key Moments - 2 Insights

Why does the center pixel value decrease after filtering?

Because gaussian_filter smooths by averaging neighbors weighted by distance, the high center value (50) blends with lower neighbors (10), lowering it to about 22.5 (see execution_table step 3).

Why do corner pixels increase after filtering?

Corner pixels start at 10 but neighbors include the higher center pixel 50, so weighted averaging raises their value to about 12.5 (see execution_table step 4).

Visual Quiz - 3 Questions

Look at the execution table, what is the approximate filtered value of the center pixel?

A 22.5

B 10

C 50

D 15.6

💡 Hint

Check execution_table row 3 where the center pixel calculation is shown.

At which step does the gaussian_filter produce the smoothed image array?

A Step 1

B Step 2

C Step 5

D Step 3

💡 Hint

Check execution_table row 2 where gaussian_filter is applied.

If sigma was increased, how would the filtered image values change?

A No change in values

B Less smoothing, values closer to original

C More smoothing, values more blended

D Filtered image becomes sharper

💡 Hint

Gaussian sigma controls smoothing strength; higher sigma means more blending.

Concept Snapshot

gaussian_filter(image, sigma)
- Smooths image by weighted averaging neighbors
- sigma controls blur amount (higher = more blur)
- Output is same shape, float values
- Useful to reduce noise or details
- Works on numpy arrays representing images

Full Transcript

We start with a small 3x3 image array with a bright center pixel. Applying gaussian_filter with sigma=1 smooths the image by averaging each pixel with neighbors weighted by distance. The center pixel value decreases because it blends with lower neighbors, while corner pixels increase slightly due to influence from the bright center. The final filtered image is a smoothed version with less sharp contrast. Increasing sigma would increase smoothing. This process helps reduce noise or details in images.