Digital Image
Processing
Presented by Dr. MECHOUEK Khaoula
2025/2026
Email: [Link]@[Link]
1
Chapters:
1. Chapter1: Vision Systems
2. Chapter 2. Overview on Images and Histograms.
3. Chapter 3. Image Filtering
4. Chapter 4. Edge Detection Methods
5. Chapter 5. Introduction to Grayscale Mathematical
Morphology
2
Chapter4:
Edge Detection Methods
3
4-1- Introduction:
Edge detection: process of identifying sharp
changes in image intensity (object boundaries).
•The chapter introduces how derivatives help detect
these changes.
•It covers gradients, finite differences, and mask-based
methods.
•It also presents main approaches like gradient
methods and the Laplacian.
4
4.2 Edge Detection
Edge detection is the process of identifying points in an
image where the intensity changes sharply. These points
usually correspond to object boundaries. It is mainly
based on computing derivatives (like the gradient) to
locate these changes.
It's crucial for tasks like object detection, image
segmentation, and feature extraction.
5
4.3 Edge Shapes
Edges can have different shapes depending on how the
intensity changes, such as step edges (sudden change),
ramp edges (gradual change), and line edges (thin or
thick lines). Understanding these shapes helps in
choosing the appropriate detection method.
6
Figure : Different types of edges a) Step Edge: The intensity of image abruptly varies
from one value to one side of the breakage to a different value on other side. b) Roof
Edge: When intensity change is not spontaneous and appears over a finite distance
usually generated by connectivity of surfaces then line edges become roof edges. c)
Ramp Edge: When the intensity change is not spontaneous and appears a limited
distance then step edges are changed to ramp edges. d) Line Edge: The intensity of
image suddenly changes values and then returns to the starting point within short
distance . 7
•Sharpening filters: Enhance edges to make the image
clearer and more detailed.
•Edge detection: Extract edges specifically to locate
object boundaries.
•Relationship: Edge detection can be seen as a step
within sharpening, while sharpening uses edge
information to improve image quality.
•Principle of sharpening: highlight transitions in
intensity (edges)
8
4.4 Unsharp masking and highboost
Unsharp Masking
•This operation is referred to as edge enhancement,
edge crispening or unsharp masking.
•To make edges slightly sharper and crisper.
•A very popular practice in industry (where the name
came).
•Subtracting a blurred version of an image from the
image itself.
9
Figure : Edge Detection steps
10
11
12
13
Example: Apply Unsharp filter on the following blurred edge
14
15
4.5 Edge detection
using 1 and 2 derivatives
16
Figure 3 Different types of Edge detection techniques
17
18
19
20
21
22
First-derivative-based
gradient filter
23
24
The gradient of an image comes from the concept of the gradient in
mathematics, and it tells you how the intensity (brightness) of the image
changes at each pixel.
Mathematical definition
𝜕𝜕𝐼𝐼 𝜕𝜕𝐼𝐼
For an image 𝐼𝐼(𝑥𝑥, 𝑦𝑦), the gradient is:∇𝐼𝐼 =
𝜕𝜕𝑥𝑥 𝜕𝜕𝑦𝑦
𝜕𝜕𝐼𝐼
• : change in horizontal direction
𝜕𝜕𝑥𝑥
𝜕𝜕𝐼𝐼
• : change in vertical direction
𝜕𝜕𝑦𝑦
2. Gradient magnitude (strength of change): ∣ ∇𝐼𝐼 ∣= 𝐼𝐼𝑥𝑥2 + 𝐼𝐼𝑦𝑦2
Large value = strong edge
𝐼𝐼𝑦𝑦
3. Gradient direction: 𝜃𝜃 = tan−1
𝐼𝐼𝑥𝑥
Direction of the edge
25
Example
If a region of the image is smooth (same gray level):
•Gradient ≈ 0
If there is a sharp transition (black → white):
26
•Gradient is large → this is an edge
27
28
29
30
31
32
33
34
35
36
37
38
Second-derivative–based
Laplacian filter
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Canny edge detector:
The Canny edge detector is one of the most powerful edge
detection methods in image processing. It was developed by John
F. Canny and is designed to detect edges accurately while
minimizing noise and false [Link] has the following steps:
1. Smoothing: Apply a Gaussian filter to reduce noise
2. Gradient computation: Calculate edge strength and direction.
3. Non-maximum suppression: Keep only thin edges (local
maxima).
4. Double threshold: Classify pixels into strong, weak, or non-
edges.
5. Hysteresis: Keep weak edges only if connected to strong
ones.
55
Fig: Edge detection using Canny method
56
57
Conclusion:
In conclusion, different edge detection methods offer trade-offs
between simplicity, sensitivity to noise, and accuracy. Basic
operators like Sobel, Prewitt, and Roberts are fast and easy to
implement but are more sensitive to noise and produce thicker
edges. Second-derivative methods like the Laplacian operator
detect fine details but amplify noise. In contrast, the Canny
edge detector provides the best overall performance by
combining noise reduction, precise localization, and edge
tracking, resulting in clean and reliable edges, though at higher
computational cost.
58
End chapter 4
59