A brief analysis of the current status and future trends of the domestic machine vision industry chain

Sorting out the domestic machine vision industry chain

Simply put, we can divide the machine vision industry chain into bottom-level developers (core components and software providers), integrators and software service providers (secondary development), and core components and software can be further divided into light sources, lenses, industrial cameras, image acquisition cards, image processing software, etc. In the current cost structure of the entire machine vision system, component and software development accounts for 80%, which is the absolute core link and value acquirer in the industry chain.

From the perspective of technical barriers:

1) Software is the main barrier, and the underlying algorithm library is the core. Currently, it is monopolized by foreign-funded enterprises. The successful ones, such as Cognex and MVTec, are mainly cultivated by decades of automation process in foreign countries. The domestic automation process is not long, and the software algorithms are still in the research and development stage, and there are not many good applications.

2) Technology at the application level is also very critical. The main thing is to master the know-how of different application environments and make adaptive products.

At present, there are four main types of market participants in the domestic machine vision industry: international integrated automation companies, international professional machine vision companies, domestic professional machine vision companies, and domestic automation equipment companies. Among them, international companies still dominate the bottom-level developer level, while domestic companies are more focused on the secondary development level with lower added value (in the form of system integration and assembly production automation special machines), and gradually try to move up to the upstream core links on this basis.

▲Cost structure of machine vision system

1. Light source

Light source is the most fully localized link. The quality of light source lies in contrast, brightness and sensitivity to position changes. The machine vision industry mainly uses  LED  light source products. There is no universal machine vision lighting equipment at present. There are personalized solutions for each specific application instance to achieve the best effect. At present, the degree of localization in the light source industry is high and the competition is fierce.

▲Enterprises involved in internal and external lighting

2. Lens

Domestic enterprises have certain competitiveness in low-end lenses, while high-end lenses are basically dependent on imports. The basic function of a lens is to achieve beam modulation and image the target on the photosensitive surface of the image sensor to complete signal transmission. Industrial lenses can be mainly divided into fixed-focus lenses, fixed-magnification lenses, telecentric lenses, and continuously variable-magnification lenses. Different lenses are applied to different industrial sites according to requirements, and the price difference is also large.

▲Domestic and foreign lens companies participating

3. Industrial cameras

Industrial cameras are mainly imported from Europe and the United States, and domestic brands are gradually replacing imports from the low-end market. Industrial cameras are the core components of industrial vision systems. Their essential function is to convert optical signals into electrical signals, which requires higher transmission power, anti-interference ability and stable imaging capabilities.

▲ Domestic and foreign industrial camera companies participating

4. Image acquisition card

Image acquisition cards are relatively well developed and mature in China. They are also called video capture cards. This component is usually a card inserted into a PC. The function of this acquisition card is to connect the camera to the PC. It obtains data (analog or digital signals ) from the camera and then converts it into information that the PC can process.

5. Image processing software

Image processing software is basically monopolized by foreign companies, and domestic companies have made some arrangements in secondary development. Industrial vision software performs various operations on digital signals to extract the characteristics of the target, and then controls the equipment actions on site according to the results of the judgment, automatically completing image acquisition, display, storage and processing. The current popular development model is "software platform + visual development kit". The development kit encapsulates various commonly used image processing algorithms based on the software platform. Software engineers can directly call the encapsulated algorithms to implement various complex image processing functions, reducing the difficulty and workload of secondary development.

▲Domestic and foreign image processing software companies

6. System Integration

Domestic manufacturers are developing rapidly in the integration sector, especially in areas where foreign capital has not yet deployed, or in non-standard automation areas such as 3C, etc. Domestic integration manufacturers have little profit margin for simple secondary development. After completing a good layout in the downstream of a certain industry, they will try to gradually extend to the upstream bottom-level development and replace the import of core software and hardware.

The most important downstream of machine vision

Machine vision is widely used in electronics and semiconductors, automobile manufacturing, food packaging, pharmaceuticals and other fields, among which electronic automobiles and electronics are currently the most important application areas of machine vision.

▲Global machine vision downstream demand structure

1. Downstream application areas - electronics

According to the data from the Prospective Research Institute, the electronics industry contributes to nearly 50% of the demand for machine vision, which is mainly used for high-precision manufacturing and quality inspection of wafer cutting, 3C surface inspection, touch screen manufacturing, AOI optical inspection, PCB printed circuits, electronic packaging, screen printing, SMT surface mounting, SPI solder paste inspection, semiconductor alignment and identification, etc. Take the iPhone as an example, its entire production process requires more than 70 systems. In the future, the demand for consumer electronics such as global smartphones, tablets and wearable devices is expected to explode.

Taking the 3C industry as an example, we believe that the demand for machine vision in the industry will continue to grow rapidly in the future. The main demand comes from several aspects: 1) Advances in visual technology (many glass and screen defect detection technologies are currently not feasible) will drive the expansion of applicable fields; 2) As domestic smartphones gradually become mid-to-high-end, the profit margins of mobile phone manufacturers will increase, and the application of visual inspection in domestic mobile phone production lines is expected to be promoted.

▲Typical application of machine vision in SMT assembly line

2. Downstream application areas - automobiles

According to the data from the Prospective Research Institute, the automotive industry contributes about 15% of the demand for machine vision, which is mainly used in the manufacturing process of almost all systems and components, such as body assembly inspection, panel printing quality inspection, character inspection, precision measurement of part dimensions, workpiece surface defect inspection, free-form surface inspection, gap inspection, etc. Currently, a production line is equipped with about a dozen machine vision systems. In the future, as the trend of automobile quality control, automobile intelligence and lightweighting puts higher requirements on inspection, the demand for machine vision technology will gradually increase.

For example, 3D vision systems can measure gaps and align each vehicle with high accuracy, and fully inspect all doors and bodies that are assembled. 3D vision systems can also help chassis manufacturers automate the loading, unloading and inspection of body panels in racks, and detect defective components on the rack before they are picked up by automated equipment, thereby reducing the need to weld defective components together.

▲Application of machine vision in vehicle body inspection

3. Downstream application areas - pharmaceuticals

According to the data from the Prospective Research Institute, the pharmaceutical industry contributes about 7% of the demand for machine vision, which is mainly used in drug bottle packaging defect detection, capsule packaging quality inspection, drug tablet detection, production date coding inspection, tablet color recognition and sorting, etc. At present, most companies have 1-2 sets of machine vision systems on their production lines, while the actual demand should be at least 5. In the future, as the automation upgrade and transformation of the pharmaceutical industry accelerates, the penetration rate will continue to increase.

For example, in the inspection process after drug packaging, machine vision can be used to quickly and accurately detect whether the object is intact. By setting the image sensor, the image information of the packaged object can be obtained, and each pill or bottle can be inspected and compared according to the pre-set area parameters. In this way, damaged pills or missing bottles will be detected and passed normally.

▲Schematic diagram of drug blister testing

▲ Diagram of bottle shortage detection

4. Downstream application areas - food

Food and packaging are also important downstream areas for machine vision applications. They are mainly used for high-speed inspection, appearance packaging inspection, food packaging defect inspection, appearance and internal quality inspection, sorting and color sorting, etc. The amount used in a single production line varies greatly among different products. At present, machine vision is widely used in large food companies (such as Yili and Mengniu), but its penetration rate in the industry as a whole is not high.