Chapter 3 Artificial Intelligence in Robots and Automation 3.

12 Image and its Fundamentals

3.12 Image and its Fundamentals

3.12.1 Definition of an Image
An image is a two-dimensional representation of a three-dimensional scene. Mathematically, it can
be defined as a two-dimensional function f (x, y), where x and y are spatial coordinates, and the
amplitude of f at any pair of coordinates (x, y) is called the intensity or gray level of the image at
that point.

3.12.2 Types of Images

• Binary Images
• Grayscale Images
• Color Images

3.12.3 Image Resolution

Image resolution refers to the detail an image holds. It’s typically described in terms of pixel reso-
lution (total number of pixels) or spatial resolution (pixels per unit).

3.13 Image Processing

3.13.1 Definition
Image processing is a method to perform some operations on an image, in order to get an enhanced
image or to extract some useful information from it.

3.13.2 Types of Image Processing

• Digital Image Processing: Digital Image Processing is a sophisticated method that involves
the manipulation of digital images using computers, where images are processed as 2D ar-
rays of discrete values. This technology encompasses various applications, including image
enhancement and restoration, pattern recognition, image compression, and machine vision
applications. One of the key advantages of digital image processing is its high precision and
accuracy in results. Additionally, it offers reproducible results and provides flexibility in ap-
plying various algorithms, making it ideal for modern computational tasks and analysis.
• Analog Image Processing: Analog Image Processing operates by processing images as con-
tinuous signals, primarily dealing with analog signals like printouts and photographs. This
traditional approach finds extensive applications in television image processing, photography
development, and radar signal processing. The distinguishing characteristics of analog im-
age processing include its continuous signal processing nature, reliance on hardware-based
processing systems, and capability for real-time processing. While less flexible than digital
methods, analog processing remains relevant in specific applications where continuous signal
handling is preferred.

78
Chapter 3 Artificial Intelligence in Robots and Automation 3.14 History of Image Processing

• Optical Image Processing: Optical Image Processing utilizes optical systems and components
for image manipulation, employing optical elements such as lenses and mirrors to process im-
ages. This method is particularly valuable in applications like holography, optical correlation,
and optical filtering. The notable features of optical image processing include its ability to
perform real-time processing without the need for digitization and its inherent parallel pro-
cessing capability. This approach proves especially effective in scenarios where direct optical
manipulation offers advantages over digital or analog alternatives.

3.13.3 Applications of Image Processing

• Medical Imaging

• Remote Sensing

• Pattern Recognition

• Video Processing

• Machine/Computer Vision

3.14 History of Image Processing

3.14.1 Early Developments
• 1920s: Invention of the Bartlane cable picture transmission system

• 1960s: Development of digital image processing techniques for satellite imagery enhancement

• 1970s: Medical imaging and satellite remote sensing drive further advancements

3.14.2 Key Milestones

• 1964: First digital image scanner

• 1969: First digital image processing techniques applied to correct distortions in lunar images

• 1973: First MRI image produced

• 1980s-1990s: Rapid development of digital cameras and image processing software

• 2000s onward: Integration with machine learning and AI technologies

3.15 Examples of Fields that Use Image Processing

Image processing has applications in a wide range of fields, revolutionizing how we analyze and
interpret visual data. Here are some key areas where image processing plays a crucial role:

79
Chapter 3 Artificial Intelligence in Robots and Automation
3.15 Examples of Fields that Use Image Processing

3.15.1 Medical Imaging

• Diagnostic Imaging: X-rays, CT scans, MRI, ultrasound

• Image-Guided Surgery: Real-time imaging for precise surgical procedures

• Microscopy: Enhancement of microscopic images for cell analysis

• Telemedicine: Remote diagnosis through digital image transmission

3.15.2 Remote Sensing and Earth Observation

• Satellite Imagery: Land use mapping, urban planning, disaster management

• Weather Forecasting: Analysis of meteorological images

• Environmental Monitoring: Tracking deforestation, pollution, and climate change

• Agriculture: Crop health monitoring and precision farming

3.15.3 Security and Surveillance

• Facial Recognition: Identification in security systems

• Biometric Systems: Fingerprint and iris recognition

• Video Surveillance: Automated monitoring and threat detection

• Forensic Analysis: Enhancement of crime scene images

3.15.4 Automotive Industry

• Advanced Driver Assistance Systems (ADAS): Lane departure warning, pedestrian detec-
tion

• Autonomous Vehicles: Object recognition and navigation

• Quality Control: Inspection of manufactured parts

3.15.5 Entertainment and Media

• Film and Video Production: Special effects, color correction

• Virtual Reality (VR) and Augmented Reality (AR): Image processing for immersive ex-
periences

• Video Games: Texture mapping, rendering, and motion capture

• Digital Photography: Image enhancement and manipulation

80
Chapter 3 Artificial 3.16
Intelligence
Differences
in Robots
between
andRobot
Automation
Vision, Machine Vision, and Computer Vision

3.15.6 Robotics and Industrial Automation

• Machine Vision: Quality control in manufacturing
• Robot Guidance: Visual navigation for industrial robots
• Defect Detection: Automated inspection systems

3.15.7 Astronomy and Space Exploration

• Telescope Image Enhancement: Improving images of distant celestial objects
• Planetary Exploration: Analysis of images from space probes and rovers
• Star and Galaxy Classification: Automated categorization of astronomical objects

3.15.8 Document Processing

• Optical Character Recognition (OCR): Converting images of text into editable documents
• Signature Verification: Automated authentication in banking and legal documents
• Document Scanning and Archiving: Digitization of paper records

3.15.9 Art Conservation and Restoration

• Digital Restoration: Recovering damaged or faded artworks
• Analysis of Painting Techniques: Studying brushstrokes and layering
• Authentication: Detecting forgeries through image analysis

3.16 Differences between Robot Vision, Machine Vision, and

Computer Vision
While these terms are often used interchangeably, there are subtle differences in their focus and
application. Understanding these distinctions is crucial for grasping the nuances of image processing
in various contexts.

3.16.1 Definitions
• Robot Vision: The ability of a robot to visually perceive and understand its environment,
typically for navigation and interaction purposes.
• Machine Vision: The use of computer vision in industrial and manufacturing applications,
often for automated inspection or analysis.
• Computer Vision: The broader field of study focused on how computers can gain high-level
understanding from digital images or videos.

81
Chapter 3 Artificial 3.16
Intelligence
Differences
in Robots
between
andRobot
Automation
Vision, Machine Vision, and Computer Vision

3.16.2 Comparative Analysis

Aspect Robot Vision Machine Vision Computer Vision

Primary Focus Real-time perception Industrial automation General image under-
and interaction and quality control standing and analysis
Typical Applications Navigation, object Defect detection, as- Object recognition,
manipulation sembly verification scene understanding
Environment Dynamic, often un- Controlled, often Varied, from con-
structured structured trolled to uncon-
trolled
Processing Speed Emphasis on real-time High-speed process- Varies based on appli-
processing ing for production cation
lines
Autonomy Level High autonomy in de- Predefined tasks and Ranges from auto-
cision making decisions mated to assistive
Hardware Integration Tightly integrated Integrated with indus- Can be software-
with robotic systems trial equipment only or hardware-
integrated

Table 3.1: Comparison of Robot Vision, Machine Vision, and Computer Vision

3.16.3 Key Differences

Robot Vision vs. Machine Vision

• Purpose: Robot vision is for guiding robot actions, while machine vision is for industrial
inspection and measurement.

• Mobility: Robot vision systems are often mobile, while machine vision systems are typically
stationary.

• Adaptability: Robot vision must adapt to changing environments, while machine vision op-
erates in more controlled settings.

Machine Vision vs. Computer Vision

• Scope: Machine vision is a subset of computer vision focused on industrial applications.

• Standardization: Machine vision often uses standardized components and protocols, while
computer vision is more diverse in its implementations.

• Accuracy Requirements: Machine vision typically requires higher accuracy for specific
tasks, while computer vision may deal with more general recognition problems.

82
Chapter 3 Artificial Intelligence in Robots and Automation 3.17 Steps in Image Processing

Computer Vision vs. Robot Vision

• Breadth: Computer vision is a broader field encompassing various applications, while robot
vision is specifically tailored for robotics.

• Action Integration: Robot vision is directly integrated with action systems, while computer
vision may or may not be linked to physical actions.

• Research Focus: Computer vision often focuses on advancing general vision algorithms,
while robot vision emphasizes practical application in robotics.

3.16.4 Overlapping Areas

Despite these differences, there is significant overlap among these fields:

• All three rely on image processing techniques.

• They share common algorithms for tasks like edge detection, pattern recognition, and 3D
reconstruction.

• Advancements in one field often benefit the others.

3.17 Steps in Image Processing

Image processing typically involves a series of steps, each contributing to the overall goal of extract-
ing useful information from images or enhancing them for specific purposes. The following outlines
the main steps in a typical image processing pipeline:

Image
Preprocessing Segmentation
Acquisition

Feature
Restoration Enhancement
Extraction

Image Template
Recognition
Analysis Matching

Figure 3.1: Steps in Image Processing

83
Chapter 3 Artificial Intelligence in Robots and Automation 3.17 Steps in Image Processing

3.17.1 1. Image Acquisition

• Definition: The process of capturing or obtaining the digital image.

• Methods:

– Digital cameras
– Scanners
– MRI, CT, or X-ray imaging devices
– Satellite or aerial imaging systems

• Considerations:

– Resolution
– Color depth
– Image format (e.g., JPEG, PNG, TIFF)

3.17.2 2. Preprocessing
• Purpose: To improve the image data by suppressing unwanted distortions or enhancing some
image features for further processing.

• Techniques:

– Noise reduction
– Contrast enhancement
– Color correction
– Geometric transformations (e.g., rotation, scaling)

3.17.3 3. Segmentation
• Definition: The process of partitioning an image into multiple segments or objects.

• Methods:

– Thresholding
– Edge detection
– Region-based segmentation
– Clustering techniques

• Goal: To simplify and/or change the representation of an image into something more mean-
ingful and easier to analyze.

84
Chapter 3 Artificial Intelligence in Robots and Automation 3.17 Steps in Image Processing

3.17.4 4. Feature Extraction

• Purpose: To extract quantifiable features or characteristics from the image.

• Types of features:

– Color features (e.g., histograms)

– Texture features
– Shape features
– Spatial relationship features

• Applications: Object recognition, pattern detection, image classification

3.17.5 5. Enhancement
• Goal: To improve the quality of the image for visual interpretation or for further analysis.

• Techniques:

– Histogram equalization
– Adaptive contrast enhancement
– Unsharp masking
– Color enhancement

3.17.6 6. Restoration
• Purpose: To recover an image that has been degraded by a known cause.

• Common degradations:

– Motion blur
– Noise
– Camera misfocus

• Techniques:

– Wiener filtering
– Inverse filtering
– Constrained least squares restoration

85
Chapter 3 Artificial Intelligence in Robots and Automation 3.17 Steps in Image Processing

3.17.7 7. Recognition
• Definition: The process of identifying objects, persons, or patterns in an image.

• Approaches:

– Statistical pattern recognition

– Structural pattern recognition
– Neural network-based recognition

• Applications: Face recognition, optical character recognition (OCR), object detection

3.17.8 8. Template Matching

• Definition: A technique for finding small parts of an image that match a template image.

• Methods:

– Cross-correlation
– Sum of squared differences
– Normalized methods

• Applications: Object detection, image registration, quality control in manufacturing

3.17.9 9. Image Analysis

• Purpose: To extract meaningful information from images.

• Techniques:

– Measurement of object properties (size, shape, color)

– Classification of image content
– Scene interpretation

• Output: Quantitative or qualitative data about the image content

86