Fundamental Steps in Digital Image Processing:
Components of an Image Processing System, Image Sampling and Quantization,
Some
Basic Relationships Between Pixels.
Components of an Image Processing System
1. Image Sensors
o This is the first stage of any image processing system.
o They capture the image from the real world (problem domain).
o Example: a digital camera uses CCD (Charge Coupled Device) or CMOS
sensors.
o The sensor converts incoming light into an electrical signal.
o A digitizer then changes this signal into digital form (0s and 1s) that the
computer can understand.
o Without sensors, no digital image can be formed.
2. Specialized Image Processing Hardware
o This hardware is specially designed for image-related tasks that need very
high speed.
o It often includes:
▪ Digitizers to convert analog signals into digital images.
▪ Arithmetic Logic Units (ALUs) to perform fast parallel operations
like addition, subtraction, or averaging of images.
▪ Graphics Processing Units (GPUs) to handle heavy matrix and deep
learning operations.
o Such hardware is sometimes called the front-end subsystem because it
handles raw image data quickly before the main computer processes it.
o Example: reducing noise by averaging video frames at 30 fps.
3. Computer
o The main controller of the image processing system.
o It can be a general-purpose computer such as a PC, workstation, or even a
supercomputer depending on the application.
o Performs most of the processing and executes algorithms through software.
o In dedicated applications, custom-designed computers may also be used for
faster results.
o In short, the computer acts as the brain of the system.
4. Image Processing Software
o Provides the actual tools and functions for processing and analyzing images.
o Includes pre-written modules for tasks like image enhancement, restoration,
segmentation, recognition, and compression.
o Example: MATLAB Image Processing Toolbox, Python libraries like
OpenCV.
o A good software system also allows users to write their own code and
combine it with existing functions.
o Thus, software provides flexibility and makes the system user-friendly.
5. Mass Storage
o Digital images take up a large amount of space, so storage is essential.
o Storage is usually divided into three levels:
▪ Short-term storage (RAM, frame buffers) → for temporary use
during processing.
▪ On-line storage (hard disks, SSDs) → for data that must be accessed
frequently.
▪ Archival storage (magnetic tapes, optical disks) → for long-term
backup of rarely used data.
o Example: A single 1024×1024 pixel image with 8-bit intensity needs about 1
MB space uncompressed.
o Hence, compression and efficient storage methods are widely used.
6. Image Displays
o Output devices for visualizing processed images.
o The most common are flat-screen color monitors driven by graphics/display
cards or GPUs.
o Advanced systems may use stereo displays or head-mounted goggles for 3D
visualization.
o Displays allow users to analyze results instantly after processing.
7. Hardcopy Devices
o Used when a permanent physical copy of the image is needed.
o Devices include laser printers, inkjet printers, film recorders, and CD/DVD
writers.
o Film gives the highest resolution, but paper printouts are more common for
documentation and reports.
o For presentations, images can be shown on projectors or transparencies.
8. Networking and Cloud
o Almost all modern systems are connected to networks or the cloud.
o This allows image data to be transmitted, shared, or stored remotely.
o The main issue is bandwidth since images and videos require high data rates.
o Broadband and optical fiber technologies have improved this significantly.
o Cloud services provide storage, remote access, and even online image
processing.
CONCLUSION:
An image processing system has input devices (sensors) to capture images, specialized
hardware + computer + software to process them, storage to save them, output devices
(displays, hardcopy) to present them, and network/cloud to share and store data remotely.
Image Digitization – Sampling and
Quantization
Digital image processing starts with converting a real-world analog image into a form that a
computer can understand.
This conversion is called Digitization, and it is performed in two steps:
1. Sampling (digitizing the coordinates)
2. Quantization (digitizing the amplitude/gray levels)
1. Sampling
● Sampling is the process of measuring the image at discrete intervals along the x-
axis (spatial coordinates).
● In images, this means dividing the image into a grid of rows and columns of pixels.
● Each pixel represents a small square portion of the image.
● The number of samples = number of pixels in the digital image.
● Higher sampling rate → more pixels → higher resolution → sharper image.
● Lower sampling rate → fewer pixels → blurry/unclear image.
● Example: A 6×6 grid gives 36 pixels, while a 256×256 grid gives 65,536 pixels
(much clearer).
Diagram: Sampling of a Signal
● The smooth curve represents a continuous signal.
● Vertical lines show how the signal is sampled at equal intervals along the x-axis.
● Each sample is stored as a point → these points form the basis of the digital version.
Meaning: Continuous data → broken into discrete points.
Diagram: Sampling in Image
● The original picture is divided into small squares (pixels).
● Left side = smooth image (high sampling).
● Right side = blocky/grainy image (low sampling).
Meaning: More pixels = better resolution, fewer pixels = poor image clarity.
Conclusion: Sampling defines the spatial resolution of the image (clarity and detail).
2. Quantization
● Quantization is the process of assigning brightness values to discrete levels along
the y-axis.
● A continuous range of intensity values is rounded off to the nearest available level.
● Example:
o If only 2 levels → image becomes black & white.
o If 8 levels → only 8 shades of gray are possible.
o If 256 levels → smooth grayscale image (standard for computers).
● Higher quantization levels → smoother shading and better quality.
● Lower quantization levels → blocky images with false contours (banding effect).
Diagram: Quantization of a Signal
● The continuous vertical values of the signal are rounded to the nearest level (0, 1, 2,
3, 4).
● This reduces infinite values to a small set of fixed levels.
Meaning: Signal amplitude is restricted to discrete steps.
Diagram: Quantization in Image
● Left image = many gray levels (smooth transitions).
● Right image = very few gray levels (blocky effect, false contours).
Meaning: More quantization levels = smoother image, fewer levels = poor quality.
Conclusion: Quantization defines the gray-level resolution of the image (brightness
details).