In simple terms, machine vision is the use of machines instead of human eyes to do measurement and judgment.

The most basic characteristic of machine vision system is to improve the flexibility and automation of production. Correspondingly, machine vision has the following main application directions and applicable scenarios.

The application of machine vision mainly has several aspects:

(1) Visual inspection can be further divided into high-precision quantitative inspection (such as cell classification in micrographs, size and position measurement of mechanical parts) and qualitative or semi-quantitative inspection without measuring instruments (such as product appearance inspection, parts identification and positioning on assembly lines, defect detection and assembly completeness detection).

(2) Robot vision: It is used to guide the operation and action of the robot in a large range, such as picking the workpiece from the chaotic workpiece pile sent by the hopper and placing it on the transmission belt or other equipment in a certain direction (namely, the hopper picking problem). As for the operation and action in a small area, it also needs the help of tactile sensing technology;

(3) Defect detection: In the process of industrial production, surface defects and impurities are prone to affect product yield. Mass application and vision can assist industrial production to identify defects efficiently and improve the yield level.

It mainly applies to the following two scenarios:

(1) Not suitable for manual operation in dangerous working environment or artificial vision is difficult to meet the requirements of the occasion;

(2) In the process of mass industrial production, the artificial vision inspection product quality efficiency is low and the accuracy is not high, and the machine vision inspection method can greatly improve the production efficiency and the degree of automation of production.

A typical industrial machine vision system includes: light source, lens (prime lens, zoom lens, telecentric lens, microscopic lens), camera (including CCD camera and COMS camera), image processing unit (or image capture card), image processing software, monitor, communication/input/output unit, etc.

Take the guided vision system of the CV-X series of Genz as an example. Its core components include:

(1) Hardware, including controller and camera. In terms of camera, the available cameras include ultra-fine 21 megapixel camera, and 3D shape measurement by using online laser profilometry (7 types of sensing heads) as measurement department;

(2) Algorithm, including multi-spectral shooting, appearance detection, intelligent learning detection, size detection, recognition character detection and other algorithms;

(3) Detection tools, including manipulator vision system, connector tools, 3D detection;

(4) Data platform, including image processing, output, etc.

From the application field of the downstream industry, machine vision has gradually extended from the industrial field to the non-industrial field. Consumer electronics, automotive and other industries have begun to widely use machine vision. From the downstream applications, due to the many advantages of machine vision in order to improve production efficiency, reduce the error in the process of production, artificial in industrial production gradually replaced by machines, machine vision applications share the industry has become the largest, one of the areas in the consumer electronics, automobile, pharmaceutical downstream industries such as manufacturing process, Machine vision system and intelligent manufacturing go hand in hand, is widely used in product size detection, defect detection, product identification, assembly positioning and other aspects.

In the non-industrial field, machine vision is mainly applied in agriculture, medical care, security, finance and transportation. Machine vision greatly strengthens the degree of agricultural automation, to achieve the separation of agricultural products, quality detection and other functions; It can be used for medical image analysis, and also has mature applications in medicine and pharmacy. It can also be used for face recognition in security and financial fields to perform identity authentication tasks; In the field of traffic can be responsible for license plate recognition and other tasks.

Why machine vision technology can be widely used in the field of industrial manufacturing production. We think there are two important reasons: (1) the reliability principle, the machine vision technology is based on the artificial intelligence architecture, its underlying support on the sensor data and core algorithm, integrates machine vision equipment automation hardware and software platform of orientation, identification, determination, for mass product standards, can achieve the high reliability; (2) Economic principle, the application of machine vision products has obvious cost advantages for the replacement of manual labor, and has higher consistency requirements.

Machine vision technology is an important branch of artificial intelligence. The research core of artificial intelligence revolves around how to make machines possess human intelligence, and its architecture can be divided into basic support layer, technical layer and application layer. The basic support layer includes big data, computing power and algorithms. The data in the basic support layer can be compared to the fuel of artificial intelligence, and the algorithm can be compared to the engine of artificial intelligence. The improvement of data volume, computing power and deep learning algorithms promote the application and development of artificial intelligence technology. The technology layer focuses on a certain aspect of human intelligence, including vision technology (machine vision, computer vision), speech technology (speech recognition, machine translation, etc.), natural language processing technology, human-computer interaction, etc. The application layer is the specific landing of artificial intelligence technology, which can be specific products and equipment (such as intelligent detection equipment) or a kind of solution (such as face recognition). From the perspective of the architecture of machine vision, it belongs to the application of artificial intelligence in the industrial field, from the underlying sensing to the integration of algorithms.

From the overseas definition, machine vision is an interdisciplinary subject, involving the comprehensive application of many fields. According to the Society of Manufacturing Engineers (SME) Chapter on Machine Vision and the Association of Robotics Industry (RIA) Chapter on Automated Vision, machine vision is a device that automatically accepts and processes an image of a real object through optical devices and contactless sensors to obtain the information needed to control the motion of a robot. Machine vision technology mainly uses multi-angle light source and sensor suitable for the object to be measured to obtain the image of the detected object, through the computer to extract information from the image, analysis, processing, and finally used for actual detection and control.

In general, machine vision is an interdisciplinary subject involving machinery, electronics, optics, automatic control, artificial intelligence, computer science, image processing and pattern recognition and many other fields.