Machine vision is a device that uses optical devices and non-contact sensors to automatically receive and process images of a real object to obtain the required information or to control the movement of a robot.
This is the definition of machine vision by the American Society of Manufacturing Engineers (SME) Machine Vision Branch and the American Robotics Industry Association (RIA) Automated Vision Branch.
▷ SME——Society of Manufacturing Engineers
▷ RIA——Robotic Industrial Association
Machine vision is the use of machines instead of human eyes for measurement and judgment. In essence, machine vision is the application of image analysis technology in factory automation. It uses optical systems, industrial digital cameras, and image processing tools to simulate human vision capabilities and make corresponding decisions. Finally, by directing a specific The device executes these decisions.
Why replace artificial vision with machine vision?
There are many reasons, the following are the main ones:
1. From the perspective of production efficiency, because the operator is easily tired under long working hours, the quality of artificial vision is low and the accuracy is not high, and machine vision can greatly improve the production efficiency and automation.
2. From the perspective of cost control, training a qualified operator requires a large amount of manpower and material resources by the enterprise manager. However, simple training is not enough, and it takes a lot of time to make the level of the operator in practice Get promoted. As long as the machine vision system is properly designed, debugged, and operated, it can be used without interruption for a long period of time, while ensuring production results.
3. In some special industrial environments, such as welding and gunpowder manufacturing, artificial vision may pose a threat to the operator's personal safety, and machine vision effectively avoids these risks to a certain extent.
What areas does machine vision cover?
A machine vision system is composed of different functional modules. Designing a successful machine vision system is very demanding on engineers.
In general, the areas of expertise covered by machine vision are as follows:
1. Electrical engineering: used for the design of hardware and software in machine vision systems.
2. Engineering Mathematics: the basis of image processing technology.
3. Physics: The basis of lighting system design.
4. Mechanical Engineering: The most widely used machine vision system. A good machine vision system can better provide manufacturing with more technical support that is conducive to improving product quality and production efficiency.
The building blocks of a machine vision system?
A complete machine vision system generally consists of optical system (light source, lens, industrial camera), image acquisition unit, image processing unit, actuator and human-machine interface and other modules. All functional modules are complementary and indispensable.
Lighting is an important factor that affects the input of machine vision systems. The design of the light source system is very important and directly related to the input data, that is, the quality of the image and the application effect.
Engineers need to first determine the effective lighting conditions and select the appropriate lighting device according to user needs and product characteristics, so as to ensure that the images generated under this lighting condition can highlight the target information characteristics required by the user.
Light sources are generally divided into visible light sources and invisible light sources. Industrially-used visible light sources include LEDs, halogen lamps, and fluorescent lamps; invisible light sources are mainly near-infrared, ultraviolet, and X-rays.
LED light source is currently the most widely used machine vision light source. It has the characteristics of high efficiency, long life, moisture resistance, shock resistance, energy saving and environmental protection. It is the best choice for engineers when designing lighting systems.
The invisible light source is mainly used to meet some specific needs, such as the inspection of pipeline welding processes, which can reach the detection point because of the invisible light penetrability.
The lens is an important component in the machine vision system, and its role is optical imaging.
The main parameters of the lens include focal length, depth of field (DOF), resolution, working distance, field of view (FOV), and so on.
Depth of field refers to the distance range of the subject before and after this optimal focus when the lens can obtain the best image.
The field of view, which represents the maximum range that the camera can observe, is usually expressed in terms of angles. Generally speaking, the larger the field of view, the larger the observation range.
Working distance refers to the distance from the lens to the subject. The longer the working distance, the higher the cost.
When designing a machine vision system, choose a lens with parameters that match the needs of the user.
Industrial cameras are essential in machine vision systems. They are like the human eye and are used to capture images. According to their different photoreceptors, cameras can be divided into: CCD cameras; CMOS cameras.
CCD—Charge Coupled Device
CMOS —Complementary MetalOxide Semiconductor
The cost of a CCD camera is higher, but the imaging quality, imaging permeability, and color richness are much better than CMOS cameras. CCD cameras can be divided into two types, line-array and area-array, according to the CCD sensor used.
The line scan camera is "line", and the image information can only be processed by the behavior unit. The resolution is high and the speed is fast. It is mainly used in industrial, medical, scientific research and other fields, supporting machine vision systems. .
Area scan cameras can obtain information about the entire image at a time, and the price is relatively cheap.
The most important component in the image acquisition unit is the image acquisition card. It is the interface between the image acquisition unit and the image processing unit. It is used to digitize the captured images and input them to the computer.
The image processing unit contains a large number of image processing algorithms. After obtaining the image, these algorithms are used to process the digital image, analyze and calculate, and output the result.
After completing the image acquisition and processing, you need to output the results of the image processing and make actions that match the results, such as rejection, alarm, etc., and display production information through the human-machine interface.
Principles of machine vision systems;
Through the optical system, the object to be captured is converted into an image signal, and the image signal is transmitted to an image acquisition card, and is converted into a digital signal according to information such as pixel distribution, brightness, and color.
The image processing unit efficiently calculates these digital signals and obtains the feature values of the shooting targets, thereby directing the equipment to perform corresponding actions according to the discrimination results.
Taking the automatic lamp inspection of foreign objects in ampoules as an example, the workflow of the machine vision system is as follows:
First, the ampoule to be inspected is transported to the inspection station by a mechanical conveying device, and the PLC sends an "object has arrived" signal.
Subsequently, the camera (camera and lens) and light source (light source) are triggered and turned on simultaneously, and the liquid in the ampoule to be inspected is acquired.
Then, the obtained state image of the medicinal solution is digitized in an image processing hardware, and the digitized image is stored in a computer.
Then, the stored information is transported to the image processing software, which processes the digital image signal and analyzes the foreign object characteristics, determines whether the liquid quality meets the requirements and makes a decision, such as good product GOOD, bad product NOTGOOD.
In the end, a control system, such as a PLC, instructs a specific device to execute the above-mentioned decision, that is, separate the good and bad products through different output channels, and display related data on the human-machine interface.