Industrial machine vision cameras and sensors are finding their way into applications both on and off the manufacturing floor. On a production line, machine vision systems far exceed the inspection capabilities of humans and can inspect thousands of parts per minute repeatedly and reliably. A machine vision system can easily assess object details that are not visible by the human eye and inspect them with greater reliability and less error.
Outside of manufacturing, many applications, including military, law enforcement, healthcare, and entertainment, are taking advantage of the same imaging technology that was used solely by industrial manufacturers.
In this primer, we explore image sensor technologies, machine vision cameras, and their capabilities to detect the visible and non-visible, beyond the capabilities of the human eye.
Inspecting the “Seen” – Visible Light
The electromagnetic spectrum is the term used to describe the entire range of light. Electromagnetic energy travels in waves and spans from very long radio waves to very short gamma rays. Light is typically measured in nanometers (nm), and each nanometer represents a wavelength of light or band of light energy. The human eye can only detect wavelengths from 400-700 nm of the part of the electromagnetic spectrum called visible light.
For the last 50 years, cameras, using either CCD (charge-coupled device) or CMOS (complementary metal oxide semiconductor) image sensors have steadily replaced human vision to deliver a more efficient, accurate and robust inspection.
Each has unique strengths and weaknesses, with advantages in different applications. For the most part, most applications have moved to use CMOS detectors; however, some applications, such as astronomy still see advantages with CCD detectors because of their low noise.
Machine vision applications built around visible light encompass virtually everything you see around you. This includes discrete part inspection for all types of containers, semiconductors, and PCBs, robotics control and guidance, surveillance and situational awareness, mail scanning and sorting, and web inspection of aluminum, steel, printed packaging, and non-woven materials.
Inspecting the “Unseen” – Infrared and X-Ray
The visible spectrum enables analysis of the surface layers of a component or an item. Infrared with wavelengths from 700 nm - 15,000 nm and X-ray with wavelengths of .01 nm – 10 nm, penetrate deeper into an object, providing information about its internal structure.
Infrared refers to the region beyond the red end of the visible color spectrum and before the microwave region. All objects radiate energy in the infrared spectrum, even objects at room temperature and frozen objects such as ice.
A wide variety of security, surveillance, military, healthcare, and manufacturing applications use infrared cameras. Their ability to detect thermally emitted IR light (heat) makes them well suited for night vision, fever detection, and an assortment of security applications, to name just a few of the growing list of uses for this versatile technology.
Let’s take a look at the different types of infrared and X-ray technologies, their applications, and the sensor technology that supports it.
Courtesy of Teledyne DALSA
Near Infrared (NIR) Imaging
Industrial image processing in the NIR spectrum is a powerful, non-intrusive technique used on production lines for quality assurance and increased productivity. This technology is proving to be a reliable manufacturing tool for inline inspection and classification of a range of products.
NIR cameras operate in the NIR spectrum with wavelengths of 750–1000 nm and use CMOS sensors, the same sensors used in visible light applications.
The printing industry uses NIR cameras to check security features woven into currency. Banknote production is becoming increasingly complex and includes a range of security features, such as embedded magnetic strips, holographs, and watermarks. These security features are difficult to inspect using visible light.
NIR cameras also inspect security labels in the pharmaceutical and cosmetic industries and is used for brand protection. Like currency, security labels have features that are only visible to NIR light, making it more difficult to produce counterfeit labels.
Precision agriculture uses NIR imaging for food and crop analysis. Healthy organic material reflects more IR light than unhealthy, dead, or inorganic matter. NIR detects moisture, fat, starch, and protein in crops, which allows farmers to make quick improvements to enhance the quality and quantity of produce, boosting revenue and profitability.
Short-wave infrared (SWIR) light is typically defined as light in the 1000 – 3000 nm wavelength range and primarily uses Indium gallium arsenide (InGaAs) infrared detectors to capture images.
SWIR can see through thick water vapors, including haze and fog. In port and harbor security scenarios where fog and haze are a regular occurrence, SWIR identifies vessels, people, and floating objects in any weather, even obscured by fog. SWIR also sees through opaque packaging such as plastic bottles to determine fill levels in pharmaceuticals, chemicals, and cosmetics for quality control.
In solar cell manufacturing, SWIR helps to maximize the efficiency of the solar cell manufacturing process. Because silicon is transparent above 1100 nm, SWIR identifies micro-cracks and anomalies in the quality of solar films and crystalline silicon.
In food production, SWIR detects water content in food. For example, when produce is deformed, water takes the place of the organic matter. Since water absorbs SWIR light, the bruised areas show up as dark spots in images with much higher contrast.
Medium-wave infrared (MWIR) use MCT (Mercury Cadmium Telluride/HgCdTe) detectors and operate in the 3000-5000 nm wavelength range. MWIR is ideal for long-distance surveillance. MWIR provides detailed images, day or night, even in challenging conditions such as humidity, fog, haze, rain, or smoke.
Another interesting application is chemical identification. Chemicals emit specific infrared signatures in terms of thermal dissipation and emissivity (a measure of an object’s ability to emit infrared energy). Law enforcement uses MWIR to detect and classify chemical compositions. It can see through bags and packages and identify chemicals and substances without touching or analyzing it. For example, if law enforcement comes across a white powdery substance, MWIR can tell them if it is fentanyl, cocaine, or baby powder.
In industrial applications, temperature and heat dynamics are key parameters that need monitoring during the manufacturing process such as welding and laser cutting, MWIR is excellent for quality control. For example, during the welding process, changes in process parameters affect the quality, size, and properties of a welded joint. Any changes that occur may compromise the mechanical behavior of the welded component, leading to a failure while using. Hence, monitoring the welding process has become vitally important to ensure the quality of welded joints. MWIR enables direct observation of the weld-arc during the process and can identify welding defects that do not meet a quality standard for production.
Long-wave infrared (LWIR) use microbolometers, made from vanadium oxide (VOx) or amorphous silicon (a-Si) detectors, and operate in the 8000-15000 nm wavelength range. LWIR detects the thermal emissions of humans, vehicles, and animals. Where MWIR is ideal for long-distance surveillance, LWIR cameras are better for perimeter surveillance and short-range surveillance. For example, a tank or Humvee with a mounted LWIR thermal imaging camera will scan to locate vehicles and enemy targets.
In manufacturing, heat-sensitive applications such as glue inspection use LWIR cameras for quality control. Hot glue inspection detects and measures the position of newly placed drops on a product. The challenge with inspecting glue is that it is completely transparent and sits typically on a dark background, so the drop is difficult to image using traditional visible imaging. Liquid glue is usually applied when hot and therefore possesses a very different thermal dissipation and emissivity than the background object, which makes it easy to detect as a clearly defined white spot in a thermal image.
One of the newer applications for LWIR is fever detection. As an alternative to thermometer guns, LWIR screens people for elevated skin temperature. Infrared cameras do not require any physical contact, mitigate the risk of user errors, and provide an almost instantaneous reading. It is not accurate enough as a diagnostic tool. However, it can identify humans with elevated skin temperatures quickly so they can be moved out of a queue for further screening. Fever scanning applications should include in the field of view, a black body radiator calibrated at a known temperature to provide a temperature reference for greater accuracy.
X-ray works in the very short-wave spectrum, .01-10 nm wavelengths, and uses Amphorous Si/Se image detectors. X-rays see through any solid surface, including human tissue, and are ideal for medical diagnostics. However, it has several other uses in manufacturing.
Multi-layered parts such as printed circuit boards use X-ray to detect problems with the inner layers. Inspecting multi-component devices such as airbags use X-ray for inspection when the product is fully assembled. Airbags have many components, including mini explosive devices that inflate the bag rapidly upon impact. Each component needs to be precisely in place for the airbag to be effective. X-ray is used for quality control before sending airbags to car manufacturers.
The detection of contaminants using X-ray technology is an essential part of quality control in food processing plants. X-ray detects metal, bone, glass, stone, and some plastic fragments, without damaging packaging for quality assurance.
In this primer, we have provided a high-level overview of imaging technologies from across the electromagnetic spectrum that take us well beyond what the eye can see. In addition to imaging, Industry 4.0, space travel, advances in genomics, healthcare, and AI are all driving the accelerated development of vision technology to meet an ever-growing number of applications well beyond manufacturing—and finding their way into everyday life. V&S