Scribd
Upload
2 views

01-introduction

The document outlines the course structure for COL783 Digital Image Analysis at IIT Delhi, including evaluation criteria and textbook information. It covers fundamental concepts of digital images, image processing techniques, and applications in various fields such as medical imaging and remote sensing. The course content includes topics like intensity transformations, image segmentation, and feature extraction.

Uploaded by

mondalmegha931
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
2 views

01-introduction

The document outlines the course structure for COL783 Digital Image Analysis at IIT Delhi, including evaluation criteria and textbook information. It covers fundamental concepts of digital images, image processing techniques, and applications in various fields such as medical imaging and remote sensing. The course content includes topics like intensity transformations, image segmentation, and feature extraction.

Uploaded by

mondalmegha931
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 35

COL783

Digital Image Analysis


Rahul Narain
[email protected]

Department of Computer Science and Engineering


Indian Institute of Technology Delhi
Course information
Evaluation (tentative):

• Assignments: 40% (4 × 10%)


• Exams:
• Minor: 20%
• Major: 30%
• Quizzes: 10%
For grading, late submission policy, audit policy, etc., see course webpage:
https://www.cse.iitd.ac.in/~narain/courses/2401-col783/
Textbook: Gonzalez and Woods, Digital Image Processing

Read the relevant section(s) after each lecture


What is an image?

image : 2D domain → intensity(ies)

• Domain is usually rectangular


• Range may be just 1 scalar (grayscale),
3 channels (colour: e.g. red, green, blue),
many (multispectral: e.g. UV, visible, IR, …)

What is a digital image?

• Domain is discrete (sampling)


• Range is discrete (quantization)
Each sample is called a pixel
Visual computing
Scienti c
visualization
Virtual/
augmented Computer
reality graphics

Computer Image
vision processing
fi
light:
shape: box [-0.5,0.5] …
intensity: rgb (10,10,10)
sphere:
center: (-0.5,0.5,-0.5)
radius: 0.5
material: mirror Vision
sphere:
center: (0.5,0.5,0.5)
radius: 0.5
material: glass
wall: Graphics
x: -1
colour: rgb (0.8,0.2,0.2)
…

Image
processing
Image processing

Image analysis

Computer vision

6 4 7
7 6 9
5 8
7 2 9 3
8 5
4 3 1 7
5 2
3 2 8
2 3 1
Applications
Medical imaging

Image enhancement Segmentation


Computed tomography

Computed tomography
Astronomy

Deconvolution

Registration, merging, enhancement


Remote sensing

Land use classi cation


fi
Object detection and recognition
Computational photography

High dynamic range


Image stitching imaging
Course content
• Intensity transformations
• Spatial ltering
• Colour image processing
• Frequency-domain ltering
• Image restoration
• Wavelet transforms
• Image compression
• Morphological operations
• Image segmentation
• Feature extraction
fi
fi
• Intensity transformations
• Spatial ltering
• Colour image processing
• Frequency-domain ltering
• Image restoration
• Wavelet transforms
• Image compression
• Morphological operations
• Image segmentation
• Feature extraction
fi
fi
• Intensity transformations
• Spatial ltering
• Colour image processing
• Frequency-domain ltering
• Image restoration
• Wavelet transforms
• Image compression
• Morphological operations
• Image segmentation
• Feature extraction
fi
fi
• Intensity transformations
• Spatial ltering
• Colour image processing
• Frequency-domain ltering
• Image restoration
• Wavelet transforms
• Image compression
• Morphological operations
• Image segmentation
• Feature extraction
fi
fi
Digital image fundamentals
The human visual system
The eye operates on a similar principle as a camera

• Iris ≈ aperture, lens = lens, retina ≈ sensor


• Retina has light-sensitive cells called cones and rods
that detect light and send signals to the brain
via the optic nerve
Cone cells

• Three types, sensitive to long, medium, short wavelengths


• Concentrated in the fovea
• Responsible for vision in bright
light (photopic vision)

Rod cells

• Only one type: no colour vision


• More sensitive to low light
(scotopic vision)
Subjective brightness is ~logarithmic in
physical light intensity

Human visual system can adapt to a huge


range of intensities (~1010)!

• But can only discriminate within a limited


range at a time
Minimum perceptible change in intensity = Weber ratio

Also varies with background illumination


Moral: Perceived brightness is not simply a function of light intensity!
Image sensing and acquisition
2D images can be acquired
even with a single sensor,
or with a sensor strip…
A simple image formation model
Assume the image is of a scene consisting of opaque objects re ecting energy from an
illumination source.

Image value f(x, y) depends on amount of illumination i(x, y) and fraction re ected r(x, y):

f(x, y) = i(x, y) r(x, y)

• Illumination component 0 ≤ i(x, y) < ∞


• Re ectance component 0 ≤ r(x, y) ≤ 1

Image value l = f(x, y) will be in the range [imin rmin, imax rmax]

• Usually scaled to standard range [0, 1] (or [0, 255], etc.)


fl
fl
fl
End of class
Homework: Read DIP book Ch. 1 and Ch. 2.1–2.3

You might also like