Machine vision is a rapidly developing branch of artificial intelligence. Simply put, machine vision uses machines to replace human eyes for measurement and judgment.
The rapid development of the internet has fueled the explosive growth of the logistics industry. Not only do annual shopping festivals like JD.com and Taobao create immense pressure and strain for logistics personnel, but even daily food delivery and supermarket deliveries are driving a surge in courier manpower. Meituan Waimai is deploying unmanned logistics, and JD.com has begun using robotic picking. Robots are bringing immense convenience to people's lives, and they are gradually becoming market darlings.
Today, robots of all types are ubiquitous around us, playing a vital role in manufacturing, transportation, and daily life. Examples include robotic mobility scooters and robotic vacuum cleaners. The technology that gives these robots "intelligent" eyes is machine vision, and thanks to the planned development of the robotics industry, the application of machine vision technology has a very broad scope.
Definition of machine vision: Machine vision is a rapidly developing branch of artificial intelligence. Simply put, machine vision uses machines to replace human eyes for measurement and judgment. It is based on biomimicry, for example, simulating the eye by acquiring images through visual sensors, followed by image processing and recognition by an image processing system.
Machine vision is mainly divided into three categories:
Monocular vision technology , which involves using a single camera to acquire images, typically only produces two-dimensional images. Monocular vision is widely used in the field of intelligent robotics. However, due to limitations in image accuracy and data stability, it needs to work in conjunction with other types of sensors, such as ultrasound and infrared sensors.
Binocular vision technology is a method that simulates how human eyes process environmental information. It uses two cameras to capture one or more images from different perspectives to establish the three-dimensional coordinates of the object being measured. Binocular vision technology can be broadly categorized into areas such as visual control of robotic arms, visual control of mobile robots, and visual control of unmanned aerial vehicles (UAVs) and unmanned surface vessels (USVs).
Multi-view vision technology refers to the use of multiple cameras to reduce blind spots and lower the probability of false detections. This technology is mainly used for measuring the motion of objects. In terms of hand-eye coordination in robotic arms, multi-view vision technology can overcome blind spots in object capture, making the robotic arm's grasping more effective. In the field of industrial robot assembly, multi-view vision can also accurately identify and locate the object being measured, thereby improving the intelligence and positioning accuracy of assembly robots.
Applications of machine vision
The main applications of machine vision are in two areas: inspection and robot vision.
1. Inspection: This can be further divided into high-precision quantitative inspection (e.g., cell classification in micrographs, measurement of the size and position of mechanical parts) and qualitative or semi-quantitative inspection without measuring instruments (e.g., product appearance inspection, component identification and positioning on assembly lines, defect detection and assembly completeness inspection).
2. Robot vision: Used to guide robot operations and actions over a large area, such as picking up workpieces from a jumbled pile of workpieces from a hopper and placing them in a specific orientation on a conveyor belt or other equipment (i.e., the hopper picking problem). For operations and actions within a small area, tactile sensing technology is still needed.
In addition, there are automated optical inspections, facial recognition, self-driving cars, product quality grading, automated inspection of printed materials, text recognition, texture recognition, and tracking and positioning. The application of machine vision technology has replaced manual sorting of express packages, greatly improving efficiency. Furthermore, machine vision technology can give robotic arms 3D vision capabilities, enabling them to guide, locate, and grasp products. Machine vision systems are also used for measurement and inspection in fields such as automotive manufacturing and medicine.
With the advent of Industry 4.0, machine vision is playing an increasingly important role in the field of intelligent manufacturing. This article explores how machine vision technology works and why it is the right choice for achieving process automation and quality improvement.
Machine vision technology is an interdisciplinary field involving artificial intelligence, neurobiology, psychophysics, computer science, image processing, pattern recognition, and many other areas. Machine vision primarily uses computers to simulate human visual functions, extracting information from images of objective objects, processing and understanding it, and ultimately using it for practical detection, measurement, and control. The biggest characteristics of machine vision technology are its high speed, large information capacity, and multiple functions.
What is machine vision?
Definition of machine vision: Machine vision is the use of optical non-contact sensing devices to automatically receive and interpret images of real-world scenes to obtain information for controlling machines or processes.
Machine vision is the automatic extraction of information from digital images for process control or inspection of manufactured products. See the example below:
Machine Vision Application Diagram
To better understand machine vision, we will use a fill level detection system used in a brewery as an example:
Figure 1. Example of beer bottle filling level inspection
As each beer bottle moves past the detection sensor, the sensor triggers the vision system to emit a strobe light and take a picture of the beer bottle.
After acquiring an image of the beer bottle and saving it to memory, the vision software will process or analyze the image and issue a pass/fail response based on the actual fill level of the beer bottle .
If the vision system detects that a beer bottle is not fully filled (i.e., it fails the inspection), it will signal a deflector to remove the bottle from the production line. Operators can view the rejected bottles and ongoing process statistics on a display screen.
In addition, machine vision systems can also measure objects, such as determining spark plug gaps or providing positional information to guide robots to align components during manufacturing and assembly. The example shown in Figure 2 mainly illustrates how machine vision systems can be used to detect whether an oil filter (right) passes or fails, and to measure the width of the central shaft head on the bracket (left).
Figure 2. Machine vision systems can perform real-time measurement and inspection on the production line, such as processing brackets (left) or oil filters (right).
In this application example, the fill level inspection system can only provide two results, which demonstrates the characteristics of a binary system:
1. If the product is qualified, the test result will be "pass".
2. If the product is not up to standard, the test result will be "failed".
What are the advantages of machine vision?
While human vision excels at qualitatively interpreting complex, unstructured scenes, machine vision, with its advantages in speed, accuracy, and repeatability, is better suited for quantitatively measuring structured scenes.
For example, on a production line, machine vision systems can inspect hundreds or even thousands of components per minute. Equipped with cameras and optics of appropriate resolution, machine vision systems can easily inspect details of objects so small that they are invisible to the human eye.
In addition, by eliminating direct contact between the inspection system and the inspected components, machine vision can also prevent component damage and avoid maintenance time and cost associated with wear and tear on mechanical parts.
By reducing human intervention in the manufacturing process, machine vision also brings additional safety and operational advantages. Furthermore, machine vision can prevent cleanroom contamination by human intervention and protect workers from hazardous environments.
Classification of machine vision systems
• Smart camera
•Based on embedded systems
• PC-based
Composition of machine vision system
• Image acquisition: light source, lens, camera, acquisition card, mechanical platform
• Image processing and analysis: Industrial control host, image processing and analysis software, graphical user interface.
• Judgment execution: Telephone unit, Mechanical unit
