Different wavelengths of light will produce different effects when used with the same lens, which is related to the lens's transmittance and chromatic aberration. Furthermore, imaging in different wavelength ranges has different applications, which is of great significance in practical project applications. For example, long-wave infrared (LWIR) imaging is widely used in non-contact temperature measurement during the COVID-19 pandemic.
lens
Lenses play a crucial role in the image quality generated by machine vision systems because they determine the sharpness of the image on the camera sensor.
Lenses can affect image quality in many ways:
1. Light transmittance is reduced due to reflection from the lens surface.
2. Spherical aberration, chromatic aberration, and defect aberrations prevent all light rays from focusing from a single point on an object onto a single point in an image.
3. Reduce light intensity towards the edges of the image.
4. Spatial distortion of the image
By choosing the right lens construction, all these effects can be minimized. Alien Eyes highlights some key considerations when selecting a lens for your specific needs, including wavelength, light transmittance, and chromatic aberration.
wavelength
When transmitting light through a lens, the wavelength of the light used is the first consideration, as it has a significant impact on chromatic aberration and light transmission. This is especially important in lens design, as infrared and ultraviolet radiation are increasingly used as alternatives to visible light to reveal information that would otherwise be invisible. While broadband light radiation is essential for certain applications (such as color imaging), using monochromatic illumination allows for optimal lens performance that is wavelength-dependent, regardless of the wavelength chosen.
Light transmittance
Generally, about 4% of the light in the air passing through a glass lens is reflected at each interface, meaning an 8% loss of incident light. Using a thin anti-reflective coating on the lens surface reduces these reflections and significantly improves light transmittance. Broadband coatings can be used for color imaging, but the best transmission comes from using a monochromatic light source and a lens coating specifically designed for that wavelength. Using wavelengths outside the coating's wavelength range actually reduces transmittance.
Color difference
When light of different wavelengths passes through the air-glass interface, each wavelength refracts a different amount. This dispersion is a measurable quantity that causes chromatic aberration in lenses, blurring the edges of objects and hindering accurate machine vision measurements. Hundreds of different types of optical glass with varying dispersion characteristics are available, allowing the production of achromatic lenses with a wide range of glass elements to compensate for chromatic aberration (Figure 1). Simpler lens designs are needed when using a monochromatic light source because there is no chromatic aberration.
Figure 1. Color difference correction
Imaging applications in different spectral bands
Line scan imaging
Line scan imaging is used to inspect continuous material and imaging objects on conveyor belts, especially when objects vary in length. Line scan (Figure 2) generates one image at a time via a line sensor of a high-speed scanning camera as the object passes underneath. Compared to area scan imaging, the resulting shorter pixel exposure time makes lens transmittance a crucial factor in ensuring sufficient light for the sensor. Sensors can provide resolutions up to 16K pixels, with 5 x 5 µm pixel monochrome line sensors exceeding 80 mm in length. Therefore, large-format lenses must be used to ensure illumination across the entire line array length without mechanical vignetting where the beam is blocked (usually by the lens mount), resulting in shadows at image edges. Anti-reflective lens coatings are essential for maximizing light throughput. However, color line scan imaging utilizes the entire visible spectrum of 400–700 nm, thus requiring achromatic lenses with broad anti-reflective coatings to eliminate chromatic aberration and improve transmittance. In a single-chip camera, the outputs of two or three adjacent lines on the sensor are combined to generate an RGB signal, while in a three- or four-sensor camera, light from the object is collected into a prism that uses the RGB components (and near-infrared (NIR) in a four-chip camera) for simultaneous detection by a single sensor. This arrangement uses a common optical path, making it more suitable for imaging 3D objects or randomly moving products without spatial displacement issues, as the R, G, B, and NIR pixels coincide for any given object location. Lenses used with prism-based cameras require special designs to compensate for the different optical paths and focusing characteristics within the camera. To avoid mechanical damage to the prism, the lens convexity must also be kept very low.
Figure 2. Line scan
Near-infrared (NIR) imaging
The near-infrared (NIR) band extends from 700 nm to 1 μm. Traditional CMOS sensors have a certain NIR sensitivity, while versions with enhanced sensitivity above 850 nm are becoming increasingly common. NIR can reveal non-surface features in foods such as fruits, vegetables, nuts, and meats, including decay, mechanical damage, and insect infestation. NIR illumination can also penetrate certain dyes and inks, allowing for product inspection through printed packaging. It also has applications in the wood, textile, paper, glass, tile, and electronics industries, and is used to inspect labels on cylindrical items (such as cans, bottles, pens, etc.) and even rotating objects. Furthermore, combining a pulsed NIR light source with an NIR pass filter can eliminate any adverse effects caused by ambient light fluctuations. If simultaneous NIR and color imaging of an object is important, using two separate cameras can present problems in obtaining accurate registration between the two images. Multi-sensor prism cameras offer a powerful alternative. Here, a prism separates the light from the sample into visible and NIR wavelength channels, each equipped with a separate sensor, thus providing identical fields of view for both images (Figure 3). The lenses used with these cameras must be well-matched to the optical properties of the prism. Here, the prism separates the light from the sample into visible light and NIR wavelength channels, each equipped with a separate sensor, thus providing identical fields of view for both images (Figure 3). The lenses used with these cameras must be well-matched to the optical properties of the prism. Here, the prism separates the light from the sample into visible light and NIR wavelength channels, each equipped with a separate sensor, thus providing identical fields of view for both images (Figure 3). The lenses used with these cameras must be well-matched to the optical properties of the prism.
Figure 3. Schematic diagram of a dual-channel color and NIR prism camera
Shortwave infrared (SWIR) imaging
The short-wave infrared (SWIR) range is approximately 1–5 µm, and many cameras are optimized to operate in the 0.9–1.7 µm range. These cameras require specialized detectors, such as indium gallium arsenide (IGaAs) or mercury cadmium telluride (MCC). SWIR cameras produce monochromatic images with resolution and detail similar to visible light, but there is a limited availability of commercially available optical glass for lenses in this band. These glasses typically offer low dispersion that decreases further with increasing wavelength, thus potentially requiring more expensive high-dispersion specialty glass elements to produce achromatic lenses. Crystalline infrared materials, such as zinc selenide and calcium fluoride, exhibit high dispersion, but their mechanical strength is far lower than glass, and they are also more expensive; therefore, most SWIR lenses continue to be made from glass with anti-reflective coatings. Water strongly absorbs light in the SWIR region, leading to numerous applications in agriculture and agronomy for detecting moisture distribution and checking the fill levels of aqueous liquids in bottles. Silicon's transparency to SWIR radiation enables the detection of non-surface semiconductors and solar cells, including impurities in semiconductor ingots and cracks in polycrystalline materials, and can differentiate plastic materials during recycling. The unique “fingerprints” created by the selective absorption of SWIR wavelengths by organic materials and plastics based on their composition allow hyperspectral imaging to become their sole identifier. This combines infrared spectroscopy with machine vision to generate color-coded images based on the chemical composition of the imaged object (Figure 4). SWIR imaging can also measure temperatures between 250°C and 800°C, making it suitable for early inspection of hot glass materials in manufacturing and welding applications.
Figure 4. The red areas in the hyperspectral image show non-surface bruising that is not visible in the color image.
Long-wave infrared (LWIR) imaging
The long-wave infrared (LWIR) band ranges from 8 to 14 μm. LWIR cameras with microbolometer sensors utilize naturally emitted radiation (thermal imaging) to achieve high-precision surface temperature distribution measurements up to 250°C. Thermal radiation can be detected at long distances, in completely dark environments, and through fog, dust, rain, and smoke. Applications include surveillance, security, and rescue, and more recently, fever screening of people during the Covid-19 pandemic (Figure 5). Industrial applications include non-destructive testing and process monitoring. LWIR optics are typically made of germanium or chalcogenide materials, such as Ge33As12Se55. Chalcogenides have similar optical and mechanical properties to germanium but different refractive indices.
Figure 5. LWIR detection of elevated human body temperature
Ultraviolet imaging
Ultraviolet (UV) imaging systems offer higher resolution than visible light, capable of resolving submicron features. This is particularly useful for detecting scratches and defects on polished or highly mirror-finished surfaces, as well as very small surface details in products such as printed circuit boards, documents, and credit cards. Because glass attenuates UV light, quartz lenses can be used for both UV-A (320–400 nm) and UV-B (290–320 nm) illumination, and achromatic doublet lenses can be constructed.
Select lens
The wavelengths used in any machine vision application are crucial for selecting the most suitable lens, with a wider spectral range requiring more complex lens designs. However, other factors must also be considered, including other defects and aberration corrections, as well as the physical environment in which the lens will be used. Given the vast array of lenses available on the market, partnering with a specialized vision technology supplier who can take a holistic approach to imaging requirements is highly beneficial. It's safe to say that few manufacturers can provide the right optical solutions for every application, so tailored, unique recommendations are the first step towards a good machine vision system.
