Morphological operations in MATLAB - Time & Space Complexity

Time Complexity: Morphological operations

O(n * m)

Understanding Time Complexity

We want to understand how the time needed to perform morphological operations changes as the image size grows.

How does the processing time increase when the image gets bigger?

Scenario Under Consideration

Analyze the time complexity of the following code snippet.

BW = imread('text.png');
se = strel('disk', 5);
BW_dilated = imdilate(BW, se);
BW_eroded = imerode(BW, se);

This code reads a binary image, creates a disk-shaped structuring element, then applies dilation and erosion morphological operations.

Identify Repeating Operations

Identify the loops, recursion, array traversals that repeat.

• Primary operation: For each pixel in the image, the structuring element is applied to check neighbors.
• How many times: This happens once for every pixel, so total operations scale with the number of pixels.

How Execution Grows With Input

As the image size grows, the number of pixels increases, so the operations increase proportionally.

Input Size (n pixels)	Approx. Operations
10 x 10 = 100	About 100 times the structuring element size
100 x 100 = 10,000	About 10,000 times the structuring element size
1000 x 1000 = 1,000,000	About 1,000,000 times the structuring element size

Pattern observation: The operations grow roughly in direct proportion to the number of pixels in the image.

Final Time Complexity

Time Complexity: O(n * m)

This means the time to complete the operation grows linearly with the number of pixels in the image and the size of the structuring element.

Common Mistake

[X] Wrong: "Morphological operations take constant time regardless of image size."

[OK] Correct: The operation must check each pixel with the structuring element, so bigger images need more work.

Interview Connect

Understanding how image size affects processing time helps you explain performance in real image processing tasks confidently.

Self-Check

"What if we use a larger structuring element? How would the time complexity change?"

Practice

1. Which MATLAB function is used to create a structuring element for morphological operations?

easy

A. imerode

B. strel

C. imopen

D. imdilate

Solution

1. Step 1: Understand structuring element creation

   The structuring element defines the shape used in morphological operations.

2. Step 2: Identify the MATLAB function for structuring elements

   In MATLAB, strel creates structuring elements like disks or lines.

3. Final Answer:

   strel -> Option B

4. Quick Check:

   Structuring element = strel [OK]

Hint: Remember: strel makes shapes for morphology [OK]

Common Mistakes:

• Confusing strel with imopen or imerode
• Using imdilate to create structuring elements
• Thinking imopen creates structuring elements

2. Which of the following is the correct syntax to perform erosion on a binary image BW using a disk structuring element of radius 3?

easy

A. BW2 = imopen(BW, strel('disk', 3));

B. BW2 = imdilate(BW, strel('disk', 3));

C. BW2 = imerode(BW, strel('disk', 3));

D. BW2 = imclose(BW, strel('disk', 3));

Solution

1. Step 1: Identify the erosion function

   In MATLAB, imerode performs erosion on images.

2. Step 2: Check the syntax for erosion with a disk structuring element

   The correct syntax is imerode(image, strel('disk', radius)).

3. Final Answer:

   BW2 = imerode(BW, strel('disk', 3)); -> Option C

4. Quick Check:

   Erosion syntax = imerode(image, strel) [OK]

Hint: imerode + strel('disk', radius) erodes image [OK]

Common Mistakes:

• Using imdilate instead of imerode
• Wrong order of arguments
• Missing quotes around 'disk'

3. What will be the output size of the binary image BW2 after applying BW2 = imdilate(BW, strel('square', 3)); if BW is a 5x5 matrix?

medium

A. Cannot determine without image content

B. 7x7

C. 3x3

D. 5x5

Solution

1. Step 1: Understand dilation size effect

   Dilation expands shapes but does not change image size in MATLAB.

2. Step 2: Check MATLAB behavior on image size

   MATLAB morphological functions keep the output image size same as input.

3. Final Answer:

   5x5 -> Option D

4. Quick Check:

   Output size = input size for imdilate [OK]

Hint: Morphology keeps image size same in MATLAB [OK]

Common Mistakes:

• Assuming dilation increases image size
• Confusing structuring element size with output size
• Thinking output depends on image content

4. Consider the code snippet:
se = strel('line', 5, 45);
BW2 = imerode(BW, se);
What is the likely error if BW is a grayscale image instead of binary?

medium

A. No error; imerode works on grayscale images

B. Error because strel cannot create line elements

C. Error because imerode expects binary images only

D. Error due to missing image conversion to binary

Solution

1. Step 1: Check imerode input types

   MATLAB's imerode supports both binary and grayscale images.

2. Step 2: Verify structuring element creation

   strel('line', 5, 45) correctly creates a line structuring element.

3. Final Answer:

   No error; imerode works on grayscale images -> Option A

4. Quick Check:

   imerode supports grayscale and binary [OK]

Hint: imerode works on grayscale too, no conversion needed [OK]

Common Mistakes:

• Thinking imerode only works on binary
• Assuming strel('line') is invalid
• Forgetting imerode supports grayscale images

5. You want to remove small noise dots from a binary image BW without changing the main shapes. Which morphological operation and structuring element combination is best?

hard

A. imopen with strel('disk', 2)

B. imclose with strel('square', 5)

C. imerode with strel('line', 3, 0)

D. imdilate with strel('diamond', 1)

Solution

1. Step 1: Understand noise removal by morphology

   Opening removes small objects (noise) while preserving shape size.

2. Step 2: Choose suitable structuring element

   A disk with radius 2 is good for removing small dots without harming main shapes.

3. Final Answer:

   imopen with strel('disk', 2) -> Option A

4. Quick Check:

   Opening + disk removes noise best [OK]

Hint: Use imopen with small disk to clean noise [OK]

Common Mistakes:

• Using imclose which fills holes
• Using erosion alone which shrinks shapes
• Using dilation alone which enlarges noise