Digital Image Fundamentals
Chapter/Topic 2
Topics in the Chapter
• 2.1 Light and electromagnetic spectrum
• 2.2 Elements of visual perception (human visual system)
• 2.3 Image sensing and acquisition
• 2.4 Image sampling and quantization
• 2.5 Mathematical tools used in digital image processing
Digital Image Fundamentals
Definition: Digital Image
• A digital image is a discrete (sampled) representation of a 2D
intensity function.
• Typically stored as a matrix/array of numbers (pixels).
• Types: grayscale, color (RGB), multispectral, binary, depth
images.
• Example: a 512×512 grayscale image with 8-bit pixels →
values in {0…255}.
Pixels = samples
Digital Image Fundamentals
Where Digital Image Processing Apply
• Acquisition → representation → enhancement/restoration → analysis → interpretation.
• Images are signals (2D), so many signal-processing ideas apply.
Key tasks: denoising, contrast enhancement, filtering, segmentation, feature extraction.
Digital Image Fundamentals
2.1
Light and Electromagnetic Spectrum
Digital Image Fundamentals
Light: Definition (in imaging)
Wave picture
• Definition: light is electromagnetic (EM) radiation that is
visible to the human eye.
λ
• In physics: EM waves can be described by wavelength (λ),
frequency (f), and energy. Photon picture
• Relationship: c = λ·f (c ≈ 3×10⁸ m/s in vacuum).
• In imaging, we care about how much light reaches the
sensor and how it varies over space.
In practice, both views are useful.
Digital Image Fundamentals
Website for reference
[Link]
+diagram
Electromagnetic Spectrum (overview)
• EM spectrum = full range of electromagnetic radiation,
ordered by wavelength or frequency.
• Bands: radio → microwave → infrared → visible →
ultraviolet → X-ray → gamma ray.
• Visible light is only a small slice, but many sensors capture
outside visible (e.g., IR cameras).
• Example: thermal cameras often sense long-wave infrared
(LWIR).
Electromagnetic spectrum diagram (NASA).
Digital Image Fundamentals
Visible Light: Wavelengths and Color
• Typical human-visible wavelengths: ~380 nm (violet) to ~700
380 nm 700 nm
nm (red).
• Shorter wavelength → higher frequency → higher photon c = λ · f
(so f = c/λ )
energy.
• Color cameras measure intensity in broad bands (R, G, B), Notation: λ (lambda) = wavelength, f =
frequency, c = speed of light
not single wavelengths.
• Example: “green” pixels integrate a range roughly around
500–600 nm depending on filters.
Digital Image Fundamentals
Radiometry vs. Photometry (very quick)
• Radiometry = measures light as energy (physics).
• Counts all electromagnetic light power (including UV/IR if your sensor
measures it).
• How much light energy is coming out?
• Photometry = measures light as humans see it (vision).
• Same light, but weighted by human eye sensitivity (humans see green
strongest, weak in red/blue, and can’t see IR/UV).
• How bright does it look to a person?
• Example 1: Infrared (IR) TV remote: A TV remote emits infrared light.
• Radiometry: detects it (because it is real light energy). |
• Photometry: almost 0, because humans can’t see IR → it doesn’t look bright.
• Example 2: Two LEDs with same power, different colors
• Suppose two LEDs have the same radiometric power (same W):
• Green LED looks very bright and Red or Blue LED looks less bright
• Why? Human eye is most sensitive near green (~555 nm).
• Radiometry says: “same power”
• Photometry says: “green appears brighter”
Digital Image Fundamentals
Reflection Types: Diffuse vs Specular
• Diffuse (Lambertian): light scatters roughly equally in
all directions.
• Specular: mirror-like reflection; depends strongly on
viewing direction.
• Real-world surfaces often have a mixture of both.
• Example: matte paper ≈ diffuse; polished metal ≈
specular.
Digital Image Fundamentals
Illumination
• Illumination means the light that falls on a surface (a scene/object).
• It is about the light coming from the source and reaching the object.
• It is not the light leaving the object toward the camera (that is called
brightness / intensity in many contexts).
• Simple example
• A table under a bright tube light → high illumination
• Same table in a dark room → low illumination
• Illumination (light falling on the object)
• Reflectance (how much the object reflects)
• A common simple relation:
White paper in dim light can look gray (low illumination)
Black cloth in bright light can still look dark (low reflectance)
Digital Image Fundamentals
2.2
Elements of Visual Perception
Digital Image Fundamentals
Eye Anatomy
• Cornea + lens: focus light onto the
retina (like a camera lens)
• Retina: the “sensor” inside the eye
• Optic nerve + brain: interpret the
signals as an image
Two important retina cells
• Rods: good for low light (night
vision), mostly black & white
• Cones: for color vision (better in
bright light)
Schematic diagram of the human eye (Wikimedia Commons).
Digital Image Fundamentals
Rods vs Cones (photoreceptors)
• Rods: very sensitive → enable night vision; mostly in the periphery.
• Cones: less sensitive but support color and high detail; concentrated in fovea.
Implication: humans see detail best where they look directly.
Digital Image Fundamentals
Brightness, Luminance, Lightness (Definitions)
• Luminance: physical measure of light leaving/arriving (photometric).
• Brightness: perceived intensity (subjective).
• Lightness: perceived reflectance (white/gray/black impression).
Why it matters: two regions with same luminance can look different
depending on context.
Digital Image Fundamentals
Brightness Adaptation
& Weber’s Law
Weber’s law (simple):
We notice a change in brightness only if the change is big enough compared to
the current brightness.
• So it’s not the absolute change that matters, but the relative change.
Example 1: In a dark room
• Background brightness (I = 10) (dim)
• A small increase (\Delta I = 2) might be noticeable. Ratio: (\Delta I/I = 2/10 = 0.2)
Example 2: In bright sunlight
• Background brightness (I = 1000) (very bright)
• The same increase (\Delta I = 2) is not noticeable.
• We may need (\Delta I = 200) to notice it. Ratio: (\Delta I/I = 200/1000 = 0.2)
Example 3:
• Adding one candle in a dark room: you notice immediately.
• Adding one candle in a room already full of lights: you barely notice.
Digital Image Fundamentals
Key Perceptions
2.3
Image Sensing and Acquisition
Digital Image Fundamentals
Image Sensing & Acquisition
Image sensing and acquisition means:
• how a camera (or sensor) “captures” a real-world scene and converts it
into a digital image (numbers).
• Sensing = the sensor detects light coming from the scene.
• Acquisition = the system records it and stores it as an image in the computer.
Outputs: grayscale (1 channel) or color (3+ channels).
Important tradeoffs: exposure, noise, resolution, dynamic range.
Digital Image Fundamentals
Steps in Image Sensing & Acquisition
Basic pipeline (like a camera)
• Light source → scene →
lens → sensor → digital
image
• Lens forms an image on the
sensor.
• Sensor (CCD/CMOS)
converts light into electrical
signals.
• ADC (Analog-to-Digital
Converter) converts the
signal into numbers.
Digital Image Fundamentals
Simple step-by-step process (camera
example)
1. Light source (sun/lamp) shines on objects
2. Objects reflect light (or emit light)
3. A lens focuses this light onto a sensor
4. The sensor (CCD/CMOS) has millions of tiny cells called pixels
5. Each pixel measures how much light hits it and makes an electrical signal
6. An ADC (Analog-to-Digital Converter) turns that signal into numbers
The numbers are stored as an image matrix
Example:
Imagine a tiny sensor of 3 × 3 pixels taking a grayscale photo.
Bright area → higher number | Dark area → lower number
Captured digital image might look like this:
Things that affect the captured image
1. Exposure time: longer → brighter but can blur if
shaky
2. Aperture: controls how much light enters
3. Noise: random grainy dots (common in low light)
4. Blur: due to motion or out-of-focus lens
Example:
A photo taken at night often looks noisy because the
sensor struggles to capture enough light.
Color Filter Array (CFA): Bayer Pattern
• Most single-sensor color cameras use a CFA: each photosite
measures one color (R, G, or B).
• Bayer mosaic is common: 50% green, 25% red, 25% blue
(e.g., RGGB).
• Demosaicing interpolates missing color values to form full
RGB pixels.
Example: green is sampled more because the eye is more
sensitive to luminance detail.
Bayer filter mosaic concept (Wikimedia Commons).
Digital Image Fundamentals
Acquisition Examples (modalities)
What is Typical
Modality
measured use
Photogra
Reflected visible phy,
Visible
light inspectio
n
Acquisition => Capturing the light X-ray
Transmitted X-ray Bones,
intensity materials
• Visible camera: records reflected visible light. Soft
NMR signal from
MRI tissue
tissues
• X-ray imaging: records transmitted X-rays (medical imaging
Heat/
Emitted IR
radiography). Thermal IR
radiation
temperat
ure
• MRI: measures signals from nuclear magnetic resonance (not
Different sensors → different physics, but all
EM illumination). produce “images” (arrays).
• Satellite remote sensing: multi/hyperspectral bands (visible +
IR).
Digital Image Fundamentals
2.4
Image Sampling and Quantization
Digital Image Fundamentals
Image Sampling and Quantization
• Sampling is the process of recording an analog
signal at regular intervals of time.
• The number of samples gives us the number of
pixels. More samples will result in higher image
quality of the digital image because of more
pixels.
• Quantization digitizes the amplitude of the
sampled signal.
• Quantization is done by rounding off the
amplitude of each sample and then assigning a
different value according to its amplitude. Each
value will represent a different color tone.
Digital Image Fundamentals
Difference Between Image Sampling and
Quantization
Digital Image Fundamentals
Advantages of Image Sampling and
Quantization
• Improved Resolution: Sampling increases the resolution of digital
images, resulting in sharper and clearer visuals.
• Color Depth Control: Quantization allows for precise control over
color depth, enabling the creation of vibrant and richly colored
images.
• Data Compression: Sampling and quantization facilitate data
compression techniques, reducing storage requirements for images
while maintaining visual quality.
• Standardization: Both processes allow for standardization of image
formats and quality, ensuring compatibility across different devices
Digitaland platforms.
Image Fundamentals
Disadvantages of Image Sampling and
Quantization
• Loss of Information: Sampling and quantization may lead to the
loss of subtle details and nuances present in the original analog
signal or image.
• Artifacts: Improper sampling and quantization can introduce
artifacts such as aliasing, banding, and color quantization errors,
degrading image quality.
• Processing Overhead: Both processes require computational
resources and processing time, especially for high-resolution images
or complex color palettes.
• Limited Dynamic Range: Quantization can result in a loss of
Digitaldynamic
Image Fundamentals range, especially in low-bit-depth images, leading to loss of
Applications of Image Sampling and
Quantization
• Digital Photography: Sampling and quantization are fundamental processes
in digital photography, enabling the capture and processing of images from
digital cameras and sensors.
• Medical Imaging: Sampling and quantization are used in medical imaging
techniques such as MRI, CT scans, and ultrasound to convert analog signals
into digital images for diagnosis and analysis.
• Remote Sensing: Image sampling and quantization are employed in satellite
imagery and remote sensing applications to capture and process data from
Earth's surface for environmental monitoring, urban planning, and agricultural
analysis.
• Graphics and Multimedia: Sampling and quantization are essential in
graphics and multimedia applications for rendering, compression, and
transmission of images in computer graphics, video streaming, and virtual
Digital Image Fundamentals
reality systems.
Revision Questions on Sampling and
Quantization
• What is 2d sampling of an image?
• 2D sampling of an image involves selecting a subset of pixels
across the image's two-dimensional space to reduce its
resolution or size while attempting to retain significant visual
information, often used in image processing and compression
techniques.
• What is image sampling frequency?
• Image sampling frequency, often referred to as pixel sampling or
pixel density, is the number of samples (pixels) taken per unit of
distance in a digital image. It determines image resolution and
quality.
Digital Image Fundamentals
Revision Questions on Sampling and
Quantization
• Why do we need image sampling?
• Image sampling is needed because it allows you to convert
continuous visual data into a digital format. This makes it easy for
computers to process and display images accurately. Sampling
defines the image resolution and detail, which affects its quality.
• What is sampling theory in image processing?
• Sampling theory in image processing is the concept of converting
a continuous image into a discrete representation by selecting and
storing values at specific sample points. It underlies digital image
formation and processing.
Digital Image Fundamentals
2.5
Mathematical Tools Used in Digital Image Processing
Digital Image Fundamentals
Image as a Matrix (Array of Values)
Grayscale image (H×W) Color image (H×W×3)
Each pixel stores brightness (often 0–255) Each pixel stores (R, G, B)
Example (3×3) Example
I = [ [10, 20, 30], Pixel at (x,y):
[40, 50, 60], (R,G,B) = (120, 200, 30)
[70, 80, 90] ]
Stacked channels:
R-channel: H×W numbers
G-channel: H×W numbers
B-channel: H×W numbers
Meaning: bigger number → brighter pixel
R G B
H×W H×W H×W
3 channels
Pixel-wise Arithmetic
Common operations Tiny numeric example
Values are clipped to [0,255] in 8-bit images
Original (2×2):
• Brighten: I′ = I + k [[ 50, 100],
[150, 200]]
• Darken: I′ = I − k
Add k = 30:
• Scale/contrast: I′ = a·I [[ 80, 130],
• Difference: D = I₁ − I₂ (change detection) [180, 230]]
Multiply a = 0.5:
[[ 25, 50],
[ 75, 100]]
Note: after operations, we often clip or rescale to valid
range (e.g., 0–255).
ntensity Transformations (Functions)
tatistics (Mean, Variance, Histogram)
eometry (Coordinates and Transformations)
Thank you