With the advent of the era of industry 4.0, the role of machine vision in the field of intelligent manufacturing is becoming more and more important, so that more users can obtain the relevant basic knowledge of machine vision, including how machine vision technology works, why it is the right choice to achieve process automation and quality improvement.

1. The present situation and trend of visual robot development in China

Robot technology is an important part of the high and new technology. Its industrialization process has just started in our country. Although we have made certain achievements, there are still many difficulties and deficiencies, so we need more attention and support from various aspects. National policy support is an important prerequisite for accelerating the industrialization of high and new technology. Chinese government departments concerned should organize forces to make a full investigation and study, and formulate a practical policy of promoting and applying robots and promoting the research and development of robots based on this. For example, the robot industry should be supported by tax, investment and loan.

Machine vision is a subject technology, widely used in production and manufacturing testing and other industrial fields, used to ensure product quality, control the production process, sense the environment and so on. Machine vision system is a target will be getting into image signal, and send it to the dedicated image processing system, according to the pixel distribution and brightness, color and other information, into a digital signal, image system of the signals for a variety of operations to extract target feature, and according to the results of determination to control the scene of the equipment.

2. Typical application of machine vision technology in the automation industry

Today, the rapid development of automation technology in our country, people's understanding of the machine vision is more profound, the view to it has also made great changes. Machine vision system improves the degree of automation of production, makes it possible to work in a dangerous environment that is not suitable for manual work, and makes mass and continuous production become a reality, which greatly improves the production efficiency and product accuracy. The performance of fast information acquisition and automatic processing also provides convenience for the information integration of industrial production. With the maturity and development of machine vision technology, it is not difficult to find that its application scope is more and more extensive. According to these fields, we can roughly summarize the five typical applications of machine vision, and the five typical applications can also basically summarize the role of machine vision technology in industrial production.

Image recognition application

Image recognition is the use of machine vision for image processing, analysis and understanding, in order to identify a variety of different patterns of targets and objects. Image recognition in the field of machine vision industry is the most typical application of two-dimensional code recognition, two-dimensional code is our usual common bar code in the most common one. A large amount of data information is stored in this small two-dimensional code, through the bar code to track the management of products. Through the machine vision system, it is convenient to recognize and read the bar code on the surface of various materials, which greatly improves the efficiency of modern production.

Image detection application

Detection is one of the most important applications in the field of machine vision industry, almost all products need to be tested, and there are many disadvantages of manual detection, manual detection accuracy is low, long time work, the accuracy is not guaranteed, and the detection speed is slow, easy to affect the efficiency of the entire production process. Therefore, the application of machine vision in image detection is also very extensive, such as the detection of coin edge characters. In the new fifth set of RMB issued in October 2000, the side of the one yuan coin was enhanced with anti-counterfeiting features, and a visual inspection system was installed at the last step of the minting process due to the strict control requirements of the production process. In addition, it can also be used in the printing process of color positioning and color check, packaging process of beverage bottle cap printing quality check, product packaging bar code and character recognition, glass bottle defect detection and so on. Among them, the machine vision system for glass bottle defect detection, also includes the scope of medicinal glass bottles, that is to say, machine vision also involves the field of medicine, its main detection includes size detection, bottle appearance defect detection, bottle shoulder defect detection, bottle mouth detection and so on.

Visual Positioning application

Visual positioning requires that the machine vision system can quickly and accurately find the parts under test and confirm their position. In the field of semiconductor packaging, the equipment needs to adjust the pick head according to the chip position information obtained by machine vision, accurately pick up the chip and bind, which is the most basic application of visual positioning in the field of machine vision industry.

Object measurement application

The biggest characteristic of machine vision industrial application is its non-contact measurement technology, which also has the performance of high accuracy and high speed, but non-contact no wear, eliminating the potential secondary damage caused by contact measurement. Common measurement applications include gear, connectors, automotive parts, IC component pins, twist drills, roded-thread testing, etc.

Object sorting applications

In fact, object sorting application is established in the recognition, detection after a link, through the machine vision system to process the image, to achieve sorting. In machine vision industrial applications, it is often used in food sorting, automatic sorting of surface defects of parts, cotton fiber sorting and so on.

Three, machine vision secondary development (industrial vision) software development comparison

At present, the more popular visual development software and visual package in the market have the following categories.

Software Platform:

VC: the most general, the most powerful function. Many users, and Windows collocation, better performance, can write their own algorithm, can also use the tool kit, and basically the tool kit supports the development of VC. It's the main platform of choice.

2 C# : relatively easy to get started, especially to complete the interface and other functions than VC+MFC difficulty is much lower, has gradually become a popular use of the platform, the algorithm in the call to the standard library or the use of C#+C++ hybrid programming. You can see that many SDK's from camera manufacturers are already using C# for their applications.

LabVIEW: NI tools graphical development platform, software development fast, especially do industrial control industry or automation testing industry of many engineers, because of the use of LabView for testing and measurement of the universality, so there are LabVIEW foundation, and then call NI Vision image tool kit development, development cycle is short, easy to maintain.

VB, Delphi: Less and less people.

Others: Java, etc. Haven't seen people use it.

Visual package:

Halcon: from MVTech, Germany. There are many underlying functional algorithms with fast operation performance, which requires certain software foundation and image processing theory to develop.

VisionPro: Image processing kit from Cognex. Performance Most algorithms perform well and are not directly comparable to Halcon in performance, but are easier to develop than Halcon.

NI Vision: The characteristics of NI are that most of the software and hardware required for automated testing have solutions, some software graphical programming, quick start, fast development cycle, the disadvantage is that not every software is very good. The advantage of the Vision toolkit is that it is much cheaper than most toolkits or algorithms at astronomical prices, and the whole kit is sold at a price rather than algorithm by algorithm, and the performance is not as fast or accurate as the first two.

MIL: A product of MaxTrox Canada, short for Matrox Imaging Library. Early promotion and popularity is good. At present, it seems that the main user is still some users of early laser equipment in use, so it is more used for positioning.

CK Vision. Chuangke company's software package, relative to the previous several tool kits for price advantage is more obvious, in addition to the machine vision needs basic functions also have, so in the domestic automation equipment, especially batch equipment at the same time need to protect the copyright of enterprises, a large amount of consumption, promotion is also good.

Meisken: Meisken's main visual products or bar code reading class, image kit has not been used, do not understand, do not evaluate.

OpenCV: I feel that OpenCV is more used in the field of computer vision, but not much in the field of machine vision, because the main applications in the field of machine vision are positioning, measurement, appearance, OCR/OCV. I feel that these are not OpenCV's specialty.