Boundary Descriptors in Image
Processing
Boundary Based Shape Analysis
• Boundary Tracking
• Procedures
• Boundary Descriptors Boundary Length Chain
Codes
• Fourier Descriptors
Image Description (Context)
• Image Description refers to extracting meaningful
information or features from the image for tasks
like recognition, classification, or retrieval.
• Two Categories:
• - Boundary Descriptors → Boundary-based
features like length, curvature.
• - Regional Descriptors → Region-based features
like area, perimeter.
• Image Representation involves converting an image into a
suitable format or representation that can be used for further
analysis or processing, Image Description refers to extracting
meaningful information or features from the image for various
tasks like object recognition, classification, or retrieval.
• Image Description
• Boundary Descriptors
• such as boundary length, diameter, curvature etc.
• Simple descriptors (length, curvature)
• Shape numbers (the first difference of smallest magnitude)
• Fourier descriptors (uses DFT for description)
• Regional Descriptors
• such as area, perimeter, compactness etc.
• Simple descriptors (Area, Perimeter, Compactness,
Circularity ratio, Rectangularity etc.)
• Topological Descriptors (measure the shape's structure
without changing by deformations)
• Texture (It provides measures of smoothness,
coarseness and regularity. There are 3 approaches:
Statistical, Structural a Spectral).
• AI responses may include mistakes. Learn more
Image Processing steps
• Image Acquisition →
Preprocessing/Enhancement → Segmentation
→ Representation/Description/Feature
Extraction →
Classification/Interpretation/Recognition →
Result
Types of Boundary Descriptors
• 1. Simple Descriptors – Boundary length,
curvature.
• 2. Shape Numbers – First difference of
smallest magnitude.
• 3. Fourier Descriptors – Uses Discrete Fourier
Transform (DFT).
FOURIER DESCRIPTORS
• Fourier descriptors (FDs) are a way to describe
the shape of an object's boundary in a way
that's robust to changes in position, size, and
rotation.
• It works by treating the object's boundary as a
signal and applying the Fourier transform.
• Imagine a shape's boundary, like the outline of
a hand
• Fourier descriptors (FDs) are a way to describe
the shape of an object's boundary in a way
that's robust to changes in position, size, and
rotation.
• It works by treating the object's boundary as a
signal and applying the Fourier transform.
• Imagine a shape's boundary, like the outline of
a hand.
STEPS IN FOURIER TRANSFORM
• Find the Boundary pixels
• Get the Boundary Coordinates
• Convert to a Complex Number Sequence
• Apply the Discrete Fourier Transform
FOURIER DESCRIPTORS
Input Bound. Bound. Co- Comp Num
Image
000 0 0 0 Pixels
0000 ord (1, Seq
0 1 1 1 0 0 1 1 1 0 1) (1,1) → 1 +
0 1 1 1 0 0 1 1 1 0
(1, 2) j1
(1, 3) (2,1) → 2 +
0 1 1 1 0 0 1 1 1 0 j1
(2, 3)
0 0 0 0 0 0 0 0 0 0 (3,1) → 3 +
(3, 3)
(3, 2) j1
(3, 1) (3,2) → 3 +
(2, 1) j2
(3,3) → 3 +
Why Complex Numbers? j3
(2,3) → they
Complex numbers are used in Fourier descriptors because 2+
j3 making
let us treat 2D boundary coordinates as a single signal,
(1,3) → 1 +
Fourier analysis simple and allowing easy handling of
j3
translation, scaling, and rotation (1,2) → 1 +
FOURIER DESCRIPTORS
Input Bound. Bound. Co- Comp Num
Image
000 0 0 0 Pixels
0000 ord (1, Seq
0 1 1 1 0 0 1 1 1 0 1) (1,1) →1+
0 1 1 1 0 0 1 1 1 0 (1, 2) j1
(1, 3) (2,1) →2+
0 1 1 1 0 0 1 1 1 0 j1
(2, 3)
0 0 0 0 0 0 0 0 0 0 (3,1) →3+
(3, 3)
(3, 2) j1
(3, 1) (3,2) →3+
(2, 1) j2
(3,3) →3+
j3
(2,3) →2+
j3
(1,3) →1+
j3
(1,2) →1+
FOURIER DESCRIPTORS
FOURIER DESCRIPTORS
FOURIER DESCRIPTORS
So the first Fourier descriptor gives us the centroid of the
shape (2,2).
FOURIER DESCRIPTORS
For u = 0 : centroid of the shape
(2,2)
For u = 1, 2,
3… 7
FOURIER DESCRIPTORS
FOURIER DESCRIPTORS
FOURIER DESCRIPTORS
FOURIER DESCRIPTORS
Boundary → Frequency Space
The sequence of boundary points (shape outline) is
converted into a set of frequency components.
Low frequencies = overall coarse shape.
High frequencies = fine details (corners, noise).
Shape Simplification
You don’t need all frequencies.
Just keep the first few descriptors (say 4–10) → this gives a
smooth, noise-free
approximation of the shape.
Simple Descriptors
• - Measure geometric features directly.
• - Examples: Boundary length, curvature.
• - Advantages: Easy to compute.
• - Limitations: Not invariant to scaling or
rotation.
Shape Numbers
• - Encodes shape as sequence of numbers.
• - First difference of smallest magnitude.
• - Helps achieve rotation invariance.
Fourier Descriptors
• - Represent boundaries in frequency domain
using Fourier Transform.
• - Compact, scale and rotation invariant.
• - Widely used for object recognition.
• Equation:
• F(k) = Σ f(n) e^(-j2πkn/N)
Comparison of Boundary Descriptors
• | Type | Pros | Cons |
• |--------------|----------------------------|----------------------
--------|
• | Simple | Easy to compute | Sensitive to
scaling/rotation|
• | Shape Numbers| Provides rotation invariance|
Less descriptive |
• | Fourier | Robust, invariant, compact | Requires
transformation |
Applications
• - Character recognition.
• - Medical image analysis.
• - Object detection in computer vision.
• - Shape-based retrieval in databases.
Conclusion
• - Boundary descriptors provide compact and
meaningful representation of shapes.
• - Choice depends on application needs:
simplicity, invariance, or robustness.
• - Fourier descriptors are most widely adopted.