The American Robotics Industry Association (RIA) Automated Visualization Division define machine vision as a system that automatically receives and processes the image of a real object through optical devices and non-contact sensors, to get the information or control the movement of the robot.
Generally speaking, machine vision replaces the human eye with a machine, but its functions are not only the reception of information but also the processing and judgment of information, like our brain.
The machine vision technology mainly uses a multi-angle light source (visible light, infrared light, X-ray, etc.) suitable for the object to be measured and a sensor to acquire an image of the detected object, and extracts information from the image through a computer for analysis and processing, and finally uses For actual detection and control. A typical machine vision system includes a light source, lens, camera, image capture card, image processing software, and so on.
To better empower the manufacturing, the requirement for the cameras and optic elements are high. It is necessary to screen the frame rate, resolution and other indicators of industrial cameras according to the requirements.
In the upstream of the machine vision industry, there are providers of light sources, lenses, industrial cameras, image acquisition cards, image processing software, etc., which are divided into hardware component production and software development. Algorithms are the core of software development while industrial cameras are the key to hardware systems, and cost accounts for about half of hardware systems.
The downstream applications of the industry are relatively wide, and the main downstream markets are the semiconductor and electronics manufacturing industries. In addition, there are applications in the auto, printing and packaging, tobacco, agriculture, medicine, and transportation sectors.
Most machine vision manufacturers in foreign countries have the advantage of industrial chain from core components (light source, lens, camera, image acquisition card, image processing software, etc.) to system integration, while Chinese manufacturers focus on machine vision system integration and equipment manufacturing.
In industrial automation production, machine vision is mainly used in three major directions: detection and measurement, guidance and positioning, identification and analysis, corresponding to different downstream application fields, such as semiconductors and electronics, automotive, food and packaging, pharmaceutical industry, etc.
According to China Industrial Information Network, in 2015, the semiconductor and electronics industries, the automotive industry, the food and packaging industry, and the pharmaceutical industry accounted for 46.6%, 10.2%, 10.1%, and 9.7% of the market, respectively.
At present, foreign manufacturers in Europe, North America, and Japan are more advanced in technology, such as Cognex and Keyence, which occupy most of the market.
China's machine vision industry chain is mainly foreign agents, system suppliers, and OEMs. At present, there are more than 300 machine vision product agents in China.
The global machine vision market is about USD 5.6 billion in 2016, and the average annual compound growth rate is about 12% in 2002-2016.
According to the Automated Imaging Association survey, the current market penetration rate of machine vision is only 5%. In the short term, the market is expected to maintain a growth rate of more than 15%, and in 2025 it will reach USD 15 billion.
The room for machine vision is spacious especially in emerging markets, however 'machine substitution' still takes a long time. Even Foxconn, which has increased its robot installation amount since 2006, has only added 10 thousand robots and 100 thousand sets of automation equipment every year. Due to various reasons, the development of robots has not been perfected, and enterprises have not stopped manual manufacturing for safety reasons. In the future, the machine vision market may grow lower than expected.