Computer Vision Implementation (CVI)

MAIN DISCUSSION CHAPTER I -- INTRODUCTION 

A. Definition 
B. Process in Computer Vision 

  1. Image Acquisition Process 
  2. Image processing process 
  3. Image data analysis (Image Analysis) 
  4. Image data understanding process (Image Understanding)

CHAPTER II -- COMPUTER VISION ARCHITECTURE 

A. Main Elements 
B. Image Processing 
C. Pattern Classification 
D. Components of a Pattern Recognition System

CHAPTER III -- RELATIONSHIP BETWEEN COMPUTER VISION AND ROBOTICS  

Computer Vision for Autonomous Robots

CHAPTER IV -- COMPUTER VISION FUNCTIONS AND IMPLEMENTATION  

  • Defense and Security Sector (Military) 
  • Autonomous Vehicles Sector Industrial Sector. 
  • Medical Image Processing Field

CHAPTER I INTRODUCTION

A. Definition

The development of information technology innovation today is very rapid, giving rise to various branches of computer science. Computer science is a systematic study of algorithmic processes that explain and transform information, whether related to theories, analysis, design, efficiency, implementation, or existing applications, one of which is computer vision.

Computer Vision is defined as a branch of science that studies how computers can recognize objects that are observed. This branch of science together with Artificial Intelligence will be able to produce a visual intelligence system (Visual Intelligence System). The difference is that Computer Vision studies more about how computers can recognize objects that are observed. Different from Computer Graphics which is more directed at digital image manipulation (visual). The simple form of Computer Graphics is 2D Computer Graphics which then developed into 3D Computer Graphics, image processing, and pattern recognition. Computer graphics are often also known as data visualization.

Computer Vision is a combination of Image Processing and Pattern Recognition. Image Processing is a field that deals with the process of image transformation. This process aims to obtain better image quality.

While Pattern Recognition, this field is related to the process of identifying objects in images or image interpretation. This process aims to extract information/messages conveyed by images.

B. Process in Computer Vision: 

To support Computer Vision tasks, there must be several supporting functions in this system, including:  

  1. Image Acquisition Process 
  2. Image processing process 
  3. Image data analysis (Image Analysis) 
  4. Image data understanding process (Image Understanding)

B.1. Image Acquisition Process 

  1. Image Acquisition in humans begins with the eyes, then visual information is translated into a format that can then be manipulated by the brain. 
  2. In line with the process above, computer vision requires an eye to capture a visual signal. 
  3. Typically the eye in computer vision is a video camera. 
  4. The camera translates a scene or image. 
  5. The output from the camera is an analog signal, where the frequency and amplitude (frequency relates to the number of signals in one second, while amplitude relates to the height of the electrical signal produced) represent the detail of the sharpness (brightness) of the scene. 
  6. The camera observes an event on one line at a time, scanning it and dividing it into hundreds of equal horizontal lines. 
  7. Each line creates an analog signal whose amplitude describes the change in 
  8. brightness along the signal line. 
  9. Then this electrical signal is converted into binary numbers which will be used by the computer for processing. 
  10. Since computers do not work with analog signals, an analog-to-digital converter (ADC) is needed to process all these signals by the computer. 
  11. This ADC will convert analog signals represented in the form of single signal information into a stream of a number of binary numbers. 
  12. These binary numbers are then stored in memory and will become raw data to be processed.

B.2. Image Processing Process 

  1. The next stage of computer vision will involve a number of initial manipulations of the binary data. 
  2. Image processing helps to improve and enhance image quality, so that it can be analyzed and processed further more efficiently. 
  3. Image processing improves the signal-to-noise ratio (s/n). 
  4. These signals are information that will represent the objects in the image. 
  5. While noise is any form of interference, lack of blur, that occurs in an object.

B.3. Image Data Analysis 

  1. Image analysis will explore the scene into the main characteristics of the object through an investigation process. 
  2. A computer program will begin to look through the binary numbers that represent visual information to identify specific features and characteristics. 
  3. More specifically, image analysis programs are used to find the edges and boundaries of objects in an image. 
  4. An edge is formed between an object and its background or between two specific objects. 
  5. These edges will be detected as a result of differences in brightness levels on different sides of one of the boundaries.

B.4. Image Data Understanding Process 

  1. This is the final step in the computer vision process, in which specific objects and their relationships are identified. 
  2. This section will involve a study of artificial intelligence techniques. 
  3. Understanding is related to template matching in a scene. 
  4. This method uses a search program and pattern matching techniques.

CHAPTER II -- COMPUTER VISION ARCHITECTURE

Computer Vision Structure
Computer Vision Structure

A. Main Elements 

Computer vision has a structure consisting of several elements, including: 

  1. Light sources are light sources used as sources for applications such as laser screens, robotics systems and so on. 
  2. Scene, is a collection of objects. 
  3. Image Device, is a tool used to change images into something that can be understood by the machine. 
  4. Image, is a picture of an object which is a representation of its actual condition. 
  5. Machine vision is a machine that interprets images related to the characteristics of patterns or objects that can be traced by the system. 
  6. Symbolic description is a system that can be used to analogize the working structure of a system to certain symbols that can be understood by the system. 
  7. Application feedback is a condition that can provide a response to receiving images from a vision system.

From the explanation above, there are three elements that underlie a vision system, namely Image Processing, Pattern Classification and Scene Analysis.

B. Image Processing 

This section functions to change or convert external images into a required representation. The following are parts of image processing:

Image Processing Phase
Image Processing Phase

C. Pattern Classification 

The idea of ​​pattern classification is how a smart machine (in this case a computer) can recognize various types and forms of patterns, such as lines, curves, shadows and various other patterns. This means that if the machine is given an input in the form of a certain pattern, the machine can understand the pattern given. The following is part of a pattern classification process:

Pattern Classification Phase
Pattern Classification Phase

Scene Analysis 

As explained previously, a relatively complex problem in computer vision is how to obtain information from an exposure (whether in the form of images or certain patterns).

Pattern Recognition 

Pattern recognition is a branch of artificial intelligence. There are several different definitions of pattern recognition, including:

  1. The assignment of a physical object or event to one or more categories. (Ward and Heart). 
  2. Science that focuses on the description and classification (recognition) of a measurement (Schalkoff). 
  3. The process of assigning a name ω to an observation x (Schürman).

Based on the definitions above, pattern recognition can be defined as a branch of artificial intelligence that focuses on methods of classifying objects into certain classes to solve certain problems.

A pattern is a composite or combination of features that are properties of an object. In classification, a pattern is a pair of variables (x, ω), where: 

  1. x is a set of observations or features (feature vector). 
  2. ω is the concept behind observation (label).

D. Pattern Recognition System Components 

The basic pattern recognition system consists of:

  1. Sensors, used to capture objects whose characteristics and features will be extracted. 
  2. Pre-processing mechanism, the mechanism for processing objects captured by the sensor is usually used to reduce the complexity of the features that will be used for the classification process. 
  3. Feature search mechanism (manual/automatic). This section is used to extract features that have gone through the pre-processing stage to separate them from features on objects that are not needed in the classification process. 
  4. Sorting algorithm, At this stage the classification process is carried out using a specific classification algorithm. The result of this stage is the classification of captured objects into predetermined criteria.

CHAPTER III -- RELATIONSHIP BETWEEN COMPUTER VISION AND ROBOTICS  

Computer Vision for Autonomous Robots 

Exploration is important and often implemented in field robotics for research purposes, as a vehicle that has autonomous exploration capabilities, it has significant potential for search, rescue, environmental monitoring, and planetary exploration operations.

This autonomous exploration capability is expected to be able to carry out Space missions, replacing the role of Telescopes due to large transmission delays. For this work, the definition of exploration problems is carried out simultaneously by covering unknown environments, mapping areas, and detecting objects of interest.

There are two challenges that must be faced in achieving exploration goals: 

  • First, we must maintain global map consistency, both in terms of distance and other measurement information, such as GPS and Magnetometer data. 
  • Second, you must be skilled at identifying important objects, whether in low light or extreme terrain, in order to minimize wasted time.

The good news is that this challenge has been completed by the 2013 NASA Sample Return Robot Challenge (NSRRC). This innovation also proves that Computer Vision is a part of AI.

CHAPTER IV -- COMPUTER VISION FUNCTIONS AND IMPLEMENTATION  

A. Computer Vision Functions 

  1. Process control, (e.g., an industrial robot or an autonomous vehicle). 
  2. Detecting events, (e.g., for visual surveillance or people counting). 
  3. Organizing information, (e.g., for indexing databases of photographs and image sequences). 
  4. Modeling objects or environments, (e.g., industrial inspection, medical image analysis or topographic models). 
  5. Interaction, (e.g., as input to devices for human-computer interaction).

B. Implementation of Computer Vision in the Defense and Security Sector (Military) 

Examples include detection of enemy soldiers or vehicles and guidance of missiles. More sophisticated systems for guidance send missiles to areas rather than specific targets and target selection is made when the missile reaches the area based on locally acquired imagery data. Modern military concepts, such as "battlefield awareness", suggest that a variety of sensors, including image sensors, provide a rich set of information about the combat scene that can be used to support strategic decisions.


For the past 10 years, MotionDSP has focused on providing image processing and Computer Vision software for military applications.

Autonomous Vehicles Field 

Examples include autonomous vehicles, which include submersibles, ground vehicles (small robots with wheels, cars or trucks), air vehicles, and unmanned aerial vehicles (UAVs). Using computer vision for navigation, i.e. to know where it is, or to generate maps of the environment (SLAM) and to detect obstacles. It can also be used to detect events, specific tasks, for example; supporting obstacle warning systems in cars,

Google Shows Off Prototype Autonomous Car
Google Shows Off Prototype Autonomous Car

Industrial Sector

Sometimes called machine vision, this information is extracted for the purpose of supporting manufacturing processes. One example is quality control where details or finished products are automatically inspected for defects.

The role of Computer Vision Software to monitor and control product quality.
The role of Computer Vision Software to monitor and control product quality.

Medical Image Processing Field

This area is characterized by the extraction of information from image data for the purpose of making a medical diagnosis of a patient. In general, image data is in the form of microscope images, X-ray images, angiography images, ultrasound images, and tomography images.

Example of computer vision techniques for characterizing finger joints in X-ray images
Example of computer vision techniques for characterizing finger joints in X-ray images

How Computer Vision Works to Analyze, Process & Understand Image Data

What is Computer Vision?

Computer Vision is the science and technology of machines that see, where "see" in this case means that the machine is able to extract information from images that is needed to complete a specific task. As a discipline, computer vision is concerned with the theory behind artificial systems that extract information from images. Image data can take many forms, such as video sequences, views from multiple cameras, or multi-dimensional data from a medical scanner. As a technological discipline, computer vision seeks to apply theory and models to the construction of computer vision systems.

Computer Vision is defined as a branch of science that studies how computers can recognize objects that are observed. This branch of science together with Artificial Intelligence will be able to produce a visual intelligence system. The difference is that Computer Vision studies more about how computers can recognize objects that are observed. However, computer graphics are more towards digital image manipulation (visual). The simplest form of computer graphics is 2D computer graphics which then develops into 3D computer graphics, image processing, and pattern recognition. Computer graphics are often also known as data visualization.

Computer Vision is a combination of Image Processing and Pattern Recognition. Image Processing is a field that deals with the process of image transformation. This process aims to obtain better image quality.

While Pattern Recognition, this field is related to the process of identifying objects in images or image interpretation. This process aims to extract information/messages conveyed by images.

1. Image Analysis

This is how computer vision does Image Analysis:

  1. Image analysis will explore the scene into the main characteristics of the object through an investigation process.
  2. A computer program will begin to look through the binary numbers that represent visual information to identify specific features and characteristics.
  3. More specifically, image analysis programs are used to find the edges and boundaries of objects in an image.
  4. An edge is formed between an object and its background or between two specific objects.
  5. These edges will be detected as a result of differences in brightness levels on different sides of one of the boundaries.

2. Image Processing

This is how computer vision processes images or pictures, also known as (Image Processing):

  1. The next stage of computer vision will involve a number of initial manipulations of the binary data.
  2. Image processing helps to improve and enhance image quality, so that it can be analyzed and processed further more efficiently.
  3. Image processing will increase the signal-to-noise ratio (s/n).
  4. These signals are information that will represent the objects in the image.
  5. While noise is any form of interference, lack of blur, that occurs in an object.

3. Image Understanding

This is how computer vision understands image data (Image Understanding):

  • This is the final step in the computer vision process, in which specific objects and their relationships are identified.
  • This section will involve a study of artificial intelligence techniques.
  • Understanding is related to template matching in a scene.
  • This method uses a search program and pattern matching techniques.

Some Examples of Computer Vision Applications:

Defense and Security Sector (Military)

The RQ-170 aircraft is an unmanned aircraft made in the United States. The control system on this aircraft is controlled by a computer located at a central military base.

The automatic lock-on missile program, a more sophisticated system for guiding missiles to areas rather than specific targets and target selection is made when the missile reaches the area based on locally acquired imagery data.

“Battlefield” awareness, suggests that multiple sensors, including image sensors, provide a rich set of information about the combat scene that can be used to support strategic decisions. In this case, automated data processing is used to reduce complexity and fuse information from multiple sensors to increase reliability.

A clear example is the detection of enemy soldiers or vehicles and the guidance of missiles. More sophisticated systems for guidance send missiles to areas rather than specific targets, and target selection is made when the missile reaches the area based on locally acquired imagery data. Modern military concepts, such as "battlefield awareness", suggest that multiple sensors, including image sensors, provide a rich set of information about the combat scene that can be used to support strategic decisions. In this case, automatic processing of data is used to reduce complexity and fuse information from multiple sensors to increase reliability.

How Does Computer Vision Capture Images?

This is how Computer Vision takes images or pictures or "Image Acquisition":

  1. Image Acquisition in humans begins with the eyes, then visual information is translated into a format that can then be manipulated by the brain.
  2. In line with the process above, computer vision requires an eye to capture a visual signal.
  3. Typically the eye in computer vision is a video camera.
  4. The camera translates a scene or image.
  5. The output from the camera is an analog signal, where the frequency and amplitude (frequency relates to the number of signals in one second, while amplitude relates to the height of the electrical signal produced) represent the detail of the sharpness (brightness) of the scene.
  6. The camera observes an event on one line at a time, scanning it and dividing it into hundreds of equal horizontal lines.
  7. Each line creates an analog signal whose amplitude describes the change in brightness along the signal line.
  8. Then this electrical signal is converted into binary numbers which will be used by the computer for processing.
  9. Since computers do not work with analog signals, an analog-to-digital converter (ADC) is needed to process all these signals by the computer.
  10. This ADC will convert analog signals represented in the form of single signal information into a stream of a number of binary numbers.
  11. These binary numbers are then stored in memory and will become raw data to be processed.

Presentation

DOWNLOAD

Reference:


Post a Comment

Previous Next

نموذج الاتصال