The most efficient image processing system, known as the vision system, is in your head. Your brain constantly identifies, processes and interprets what you see. What seems natural and easy to you is very difficult to translate into machines. All constructed vision systems strive to be as good and efficient in their operation as the human brain. Although it is still impossible to create a vision system as universal as the human brain, in specific, specific applications, the machines already allow for greater efficiency of vision control, and also ensure greater stability of the system. The vision system, unlike the human brain, will not lose concentration or get tired after a few minutes. A properly constructed vision system is a key element of the 4.0 industry production line. It ensures high quality of manufactured products in virtually every industry. The vision system will work both in the production of food and metal, it can check the presence of elements or perform accurate dimensional analysis. 3D measurements and character recognition? Nowadays, it is more a matter of planning the implementation date than developing the technology.
What does the vision system consist of?
The main element of the vision system is a camera with a lens. Why will these two elements be discussed together? Because they form an integral whole, just like the human eye. The lens corresponds to the cornea, while the camera, and more precisely the heart of the camera, i.e. the photosensitive sensor – the macula.
The light-sensitive sensor transforms the incident image into an electrical signal proportional to the light intensity. In most vision systems, monochrome (black and white) cameras are a sufficient and better solution. They provide better resolution and sharpness of the image in relation to color cameras. However, color cameras will work where color analysis is important (although this also depends on the application, because this can also be done with a monochrome camera used in the set with optical filters). Because the matrix only registers the intensity (brightness) of the incident light, to get a colorful image on the camera sensor a matrix of color filters (Bayer filters) is installed: green, red and blue, which cut off the remaining wavelengths, therefore one pixel registers the intensity of only one component of visible light . Based on the assembly of three components, the resulting color image is created. You can read more about this in our CCD / CMOS article.
Two different types of cameras are used in the construction of vision systems:
· Image cameras – they have a rectangular matrix with side up to several thousand pixels, they are used to record entire 2D images during one exposure cycle. Such cameras will be useful in vision systems, where the entire object or feature must be observed and interpreted at once.
· Line cameras – they have a matrix in the shape of a thin line with a width of several pixels, image acquisition using line cameras runs line by line, only after folding the line the whole image of the object is obtained. Line cameras will work where objects move, e.g. on a conveyor belt, allowing higher speed and inspection resolution.
· 3D cameras – mainly vision cameras using Time-of-Flight technology are used in vision systems. Such cameras send a modulated infrared beam towards the object and examine the time it takes for the beam to travel the camera-object-camera distance. Based on this, the cloud of scene points located in front of the camera is determined.
The lens plays two important roles in the vision system: it focuses the rays of light from the object’s surface on the matrix plane and limits the amount of light falling on the matrix. Today lenses are complicated constructions, often with a price equal to the camera. It is especially difficult to make large lenses with high accuracy. Therefore, along with the size of the lens, its price increases very quickly.
Depending on the way the rays are conducted and the functions offered, the lenses can be divided into several basic types:
· Fixed focal length lenses – the most popular solution used in vision systems. As the name implies, the lenses have a fixed focal length, i.e. they do not offer a change in the field of view without changing the working distance. These lenses have two adjustment rings: the first for focusing, the second for controlling the aperture. Most often, vision systems are constructed in such a way that there is no need to change the working distance, therefore fixed focal length lenses are an ideal compromise between functions and price.
· Zoom lenses – the principle of operation of zoom lenses is similar to that of fixed focal length lenses, except that they also offer the option of changing the focal length, i.e. modifying the field of view of the camera. Such lenses are more often used e.g. in robotics, especially if they are equipped with electronic control. They will also be a good solution for testing various solutions in the construction of a vision system.
· Telecentric lenses – are characterized by guiding rays parallel to the optical axis. Due to the lack of perspective error, telecentric lenses are used for precise dimensional analysis. Telecentric lenses can work with both image and line cameras. Since the field of view of telemetry lenses is equal to the diameter of the front lens of the lens, telecentric lenses are more expensive solutions with the two solutions presented above.
· 360 optics – otherwise known as pericentric lenses. The special design of these lenses allows simultaneous observation of the front and side surfaces of the lens. Depending on the model, the lenses allow, for example, to inspect the inside of the bottle cap as well as the thread with one image camera. Other solutions allow you to view the entire label placed on the neck of the bottle.
The lighting used in vision systems has been completely dominated by LEDs. LED illuminators are very efficient sources of continuous light, available in many different colors of the visible spectrum, UV and IR. LEDs are an energy-saving solution, thanks to which they do not emit much heat compared to other light sources.
The following types of illuminators are used in vision systems:
· Front-lighting – they usually have a rectangular shape with different proportions of sides: from a thin rectangle to large squares. These illuminators are used for front lighting of objects, from the same side where the camera is set. Front illuminators often have lenses on LEDs that allow focusing light over greater distances.
· Rear illuminators – the shape and characteristics of lighting resemble front illuminators, but in contrast to them they almost always have a matte, diffusing light diffuser, ensuring high homogeneity of the generated light. Rear illuminators are placed behind the object (often directly behind it) and emit light towards the camera. This type of light enhances the edges of objects.
· Ring illuminators – have LEDs arranged on a ring plan. Ring illuminators are often mounted on the camera or near the camera so that the cameras look through their opening. Ring illuminators are also used for front lighting of objects, they provide even illumination without the indicated direction.
· Shadowless lighting – especially useful when lighting highly reflective surfaces. Shadowless lighting provides smooth, homogeneous lighting that evenly emphasizes the features of objects. Shadowless illuminators can have the shape of a dome or tunnel with a hole for the camera (LEDs located in the base of the housing, dome / tunnel covered with a rough layer evenly directing light towards the object) or a flat front illuminator with a hole for the camera (dome with a radius ~ infinity, principle of operation similar to front illuminators, except that the centrally placed camera does not create shadows)
· Coaxial illuminators – they usually have a rectangular shape with LEDs located in the base and a light splitting plate that allows the light source and camera to overlap. This combination allows you to see unevenness on flat surfaces. Similar to shadowless illuminators, they are a suitable solution for vision systems in which shiny objects are observed
· Linear illuminators – illuminators dedicated to work in vision systems using line cameras. They are characterized by a very bright focused, longitudinal beam of light precisely illuminating the fragment of the object observed by the camera. Linear illuminators have a narrow strip of LEDs most often connected to a cylindrical lens so that the light can remain focused at greater working distances.
· Spotlights – small illuminators consisting of one / several LEDs. They will work where there is a need to illuminate small areas or e.g. as a light source in telecentric lenses with separate coaxial lighting
· Hyperspectral / multispectral illuminators – these illuminators are characterized by the emission of at least several different wavelengths of light. These illuminators may take the shape of the various illuminators listed above. They are most often used in tandem with hyperspectral cameras.
· Light sources – they emit strong light with specific characteristics, which is then guided by optical fiber to the inspection site. This type of illuminators is used, among others in microscopes or special illuminators, without electronics.
Vision system in reality
Vision systems are widely used in the processing and production of plastic products. Today, virtually every film production facility uses an automatic inspection system. Especially line cameras allow for efficient and detailed control of film rolled up on rolls. Other inspected materials are e.g. PET flakes, especially for optical sorters and small plastic components, such as ampoules, containers, preforms, and whole bottles.
The food industry also benefits from the use of vision systems. Both food and food packaging can be subjected to visual inspection. Vision systems can, for example, assess the freshness of meat, fruit, vegetables, search for “sugar spots” on potatoes, detect contaminants in the input feed material for food production. Packaging control can include seal quality control, OCR / OCV – label printing control, vision systems checking threads on bottles/caps using pericentric lenses, analysis of completeness of product arrangement in cartons or on pallets.
Vision systems are also present in printers, where they are used to automatically recognize and verify text, as well as assess print quality. Printing preview systems are also often used in printers, which are sent to the operator for evaluation.