SWIR Camera Lens for High-Speed Conveyor Inspection: Exposure, Motion Blur, FOV, Object Sampling and Inspection Throughput

A SWIR inspection system that performs well on a stationary sample can fail completely when the same product moves through a production line at several metres per second. High-speed conveyor inspection introduces a tightly connected group of engineering constraints: exposure time, motion blur, field of view, pixels per object, SWIR signal level, illumination intensity, frame acquisition, object spacing, processing latency and reject timing. Improving one parameter can easily make another worse. A longer exposure produces more signal but increases motion blur. A wider field of view inspects more conveyor width but reduces the pixels available for each object. A smaller aperture can improve depth of field but demands more illumination or a longer exposure. Higher conveyor speed increases throughput but reduces the time available for acquisition, classification and physical rejection.

This interaction is particularly important in SWIR because useful inspection information may come from subtle wavelength-dependent differences rather than obvious visible contrast. A material classifier, moisture detector or hidden-defect algorithm can lose accuracy long before an operator considers the image obviously blurred. High-speed SWIR machine vision therefore has to preserve the spectral and spatial integrity of each object while the production line continues moving.

The Kyptec Automation® SWIR Camera Lens collection currently provides 8.5 mm, 12.5 mm, 25 mm, 35 mm and 50 mm focal lengths within a dedicated 900–1700 nm optical family. The live product range is specified around 2 MP resolution, 2/3-inch sensor format, F1.4 aperture and C-Mount, and is positioned for industrial machine vision, food processing, pharmaceutical, electronics, textile and other automated inspection environments. For high-speed conveyor OEMs, this range is useful because it allows inspection width and object sampling to be changed without moving outside the same specialized SWIR Camera Lens platform.

Conveyor Speed Must Be Converted Into Image Motion

The most important exposure calculation is simple:

Motion during exposure = Conveyor speed × Exposure time

If a product travels at 2 m/s, that is 2000 mm/s. At an exposure of 500 µs:

2000 mm/s × 0.0005 s = 1 mm of motion

If the system resolves 0.2 mm per pixel at the object plane, the product moves approximately five pixels while the image is being exposed.

That is severe blur for many inspection tasks.

At 100 µs, movement falls to:

2000 × 0.0001 = 0.2 mm

or approximately one pixel.

At 50 µs, movement becomes 0.1 mm, or half a pixel.

A common machine-vision design approach is to constrain movement during exposure to approximately a fraction of one pixel when accurate edge or small-feature inspection is required. Machine-vision guidance likewise emphasizes that increasing exposure makes an image brighter but also increases motion blur, with permissible exposure derived from object velocity and allowable pixel displacement.

The correct allowable movement depends on the application. Whole-object material classification may tolerate somewhat more blur than inspection of a 0.5 mm contamination spot. The engineering target should therefore be defined from the smallest feature or spatial material region that must remain distinguishable.

Object-Side Pixel Size Connects FOV With Motion Blur

Exposure time cannot be selected correctly without first knowing object-side sampling.

If a camera has 1600 horizontal pixels and the horizontal field of view is 400 mm:

400 mm ÷ 1600 pixels = 0.25 mm/pixel

If allowable motion is half a pixel, the product may move approximately 0.125 mm during exposure.

For a conveyor moving at 2500 mm/s:

Maximum exposure ≈ 0.125 / 2500 = 0.00005 s = 50 µs

Now consider the same camera covering 800 mm:

800 ÷ 1600 = 0.50 mm/pixel

Half-pixel allowable motion becomes 0.25 mm, permitting approximately 100 µs exposure at the same speed.

At first glance, the wider FOV appears beneficial because a longer exposure becomes allowable in pixel terms. However, the smallest defect also occupies half as many pixels.

This demonstrates why high-speed imaging cannot optimize exposure independently from FOV.

A Wider FOV Increases Throughput but Reduces Information Per Object

OEMs often want one camera to inspect as much conveyor width as possible because wider coverage can reduce camera count and mechanical complexity.

The trade-off is spatial sampling.

Suppose a 6 mm object crosses a conveyor. At 0.25 mm/pixel it occupies around 24 pixels across its width. At 0.50 mm/pixel it occupies only around 12 pixels. A 1 mm contaminant inside that object drops from roughly four pixels to only two.

If the classifier uses average spectral information over a large object, 12 pixels may still be adequate. If the machine needs to find a small contamination region, it may not be.

The correct FOV is therefore not the widest field that fits on the sensor. It is the widest field that preserves enough pixels on the smallest production-relevant feature.

High-Speed SWIR Inspection Is Often Photon-Limited

Short exposure solves motion blur but creates another problem: fewer photons reach the sensor.

Machine-vision exposure guidance consistently identifies the same trade-off: high-speed objects require short exposure, but maintaining good signal-to-noise ratio then requires more optical signal through illumination and optics.

This becomes especially important when the inspection uses narrow SWIR wavelength bands, when the material is strongly absorbing, or when illumination is positioned in transmission geometry.

A classifier may still produce a numerical result from a dark noisy image, but the difference between good and defective material can collapse as shot noise and sensor noise become comparable to the spectral contrast being measured.

The design target should therefore be minimum exposure with sufficient signal-to-noise, not minimum exposure at any cost.

Why F1.4 Is Valuable on High-Speed SWIR Conveyors

The current Kyptec Automation® SWIR Camera Lens family provides an F1.4 maximum aperture across the 8.5 mm, 12.5 mm, 25 mm, 35 mm and 50 mm focal-length options.

A relatively wide aperture provides useful optical throughput when exposure must be shortened to tens or hundreds of microseconds.

This can be particularly valuable for fast material sorting, food inspection, textile classification, moisture inspection and industrial product differentiation, where spectral contrast must remain measurable despite motion.

However, F1.4 should not automatically become the production aperture. Opening the lens increases light collection but reduces depth of field. If objects bounce, vary in height or travel on an uneven conveyor, a slightly smaller aperture may produce more reliable focus even though it requires stronger illumination.

The optimum setting is therefore the aperture that maximizes production classification margin across expected object height, not necessarily the brightest image.

Motion Blur Should Be Specified in Pixels, Not Only Millimetres

Saying that an object moves 0.2 mm during exposure is incomplete.

Whether 0.2 mm is acceptable depends on object sampling.

At 0.05 mm/pixel, 0.2 mm represents four pixels of motion.

At 0.4 mm/pixel, it represents only half a pixel.

A stronger machine specification therefore states:

maximum allowable motion = X pixels during exposure

and then converts that requirement into microseconds using conveyor speed and object-space pixel size.

This method scales naturally if the FOV, sensor or machine speed changes later.

High-Speed SWIR Classification Can Fail Before the Image Looks Blurry

Material classification does not always depend on crisp geometric edges, so an operator may assume a slightly blurred image is harmless.

That can be misleading.

If one object smears into the conveyor background, its edge pixels become mixed. If two adjacent objects smear into one another, spectral signatures can become contaminated. A small foreign material patch can be averaged with the surrounding accepted product.

The result is not merely visual softness. It is spectral mixing.

This is why motion blur should be controlled even when the machine's primary output is material identity rather than dimensional measurement.

Object Spacing Controls More Than Mechanical Throughput

Objects moving on a conveyor should remain sufficiently separated that the inspection system can segment them reliably.

If two items overlap or touch, the image-processing system may create one region containing multiple materials.

At high conveyor speeds, longitudinal spacing also affects trigger timing and downstream rejection.

Suppose packages are 100 mm long and separated by only 20 mm at 3 m/s. The 120 mm pitch passes one object position approximately every:

120 mm ÷ 3000 mm/s = 0.040 s

or 40 milliseconds.

The entire acquisition, classification, tracking and reject scheduling system must therefore handle approximately 25 object events per second on that lane.

If several lanes are present, total decision throughput increases further.

Inspection Throughput Is Not the Same as Camera Frame Rate

A camera operating at 100 frames per second does not automatically inspect 100 products per second.

Real throughput depends on object spacing, number of objects per frame, whether each object requires several spectral bands, processing time, communication latency and rejection logic.

Conversely, one frame can contain many separated products, allowing object throughput to exceed frame rate.

For a multi-lane sorter, the relevant metric is therefore:

objects inspected correctly per second at the required defect sensitivity

rather than nominal frames per second alone.

High-speed industrial SWIR and hyperspectral systems are specifically designed around this relationship between line rate, conveyor speed and material classification throughput. Current industrial SWIR systems can operate at hundreds or thousands of lines per second depending on architecture, illustrating how tightly acquisition speed is connected to inline sorting.

Frame Interval Must Be Compared With Object Movement

At 100 frames per second, one frame begins every 10 ms.

A conveyor moving at 2 m/s travels:

2000 mm/s × 0.010 s = 20 mm

between frames.

If a product is only 12 mm long in the direction of travel, it could move through a particular inspection position between consecutive frames unless the acquisition is appropriately triggered or the FOV extends far enough in the travel direction.

This is why high-speed system design must distinguish exposure duration from frame interval.

Exposure controls blur.

Frame interval controls temporal sampling.

Both must match the conveyor speed.

Triggered Acquisition Is Usually Better Than Uncontrolled Free Running

Free-running acquisition captures images continuously without regard to object location.

This may work for continuous web inspection but can waste bandwidth or capture products at inconsistent positions when individual parts are widely spaced.

Triggered acquisition allows the system to expose when an object reaches a known point in the inspection zone.

For discrete products, this can improve positional consistency and simplify region-of-interest processing.

Trigger delay should account for the physical distance between the detection sensor and optical inspection line.

At high speed, even one millisecond matters. At 3 m/s, a one-millisecond timing error corresponds to:

3000 mm/s × 0.001 s = 3 mm

of positional change.

Kyptec Automation® KL-1408 for Wide Conveyor Inspection

The Kyptec Automation® KL-1408 8.5 MM SWIR Camera Lens provides the broadest focal-length geometry in the current Kyptec Automation® SWIR portfolio.

This makes it particularly useful to evaluate where a large belt width, multiple lanes or a broad product stream must fit inside the camera's field at limited working distance.

Applications can include large-object sorting, wide material streams and inspection systems where throughput benefits from observing several objects simultaneously.

The OEM should nevertheless calculate pixels per smallest object before choosing the widest model. Wide coverage improves scene capacity only if material classification remains adequately sampled.

Kyptec Automation® KL-1410 for Balancing Conveyor Width and Object Sampling

The Kyptec Automation® KL-1410 12.5 MM SWIR Camera Lens can provide a particularly useful compromise for high-speed systems. Its current specifications include 12.5 mm focal length, 900–1700 nm operation, F1.4 aperture, 2 MP, 2/3-inch format and C-Mount.

Where an 8.5 mm lens captures excessive empty belt area, moving to 12.5 mm can allocate more pixels to each product without forcing the inspection into a narrow-detail geometry.

For many conveyor OEMs, this balance between coverage, exposure and pixels per object is more important than simply maximizing or minimizing focal length.

Kyptec Automation® KL-1412 for Smaller Objects and Narrower High-Speed Lanes

A narrower sorting lane or smaller object can justify the Kyptec Automation® KL-1412 25 MM SWIR Camera Lens.

The tighter geometry allows a larger fraction of the sensor to be assigned to each inspected product.

That can improve segmentation of small pieces, spatial isolation of contamination and extraction of cleaner material regions.

This does not necessarily reduce throughput. If the machine uses several narrow lanes or multiple cameras in parallel, it may achieve higher total classification reliability than one extremely wide field attempting to inspect everything simultaneously.

35 mm and 50 mm Lenses Can Support High-Speed Local Inspection

High conveyor speed does not always mean wide conveyor coverage.

Some production lines move small components rapidly through a narrow inspection channel. In those situations, longer focal lengths can be appropriate.

The Kyptec Automation® KL-1414 35 MM SWIR Camera Lens and Kyptec Automation® KL-1416 50 MM SWIR Camera Lens provide tighter fields while retaining the common 900–1700 nm, F1.4, 2 MP, 2/3-inch and C-Mount architecture.

These focal lengths can be useful where high-speed inspection focuses on a controlled component region, localized material difference or narrow product lane and mechanical stand-off must remain relatively large.

Conveyor Height Variation Changes Both Focus and Sampling

Objects do not always travel at the nominal belt plane.

Food pieces, packages, recycled materials and irregular industrial parts can have significant height variation. As an object approaches the lens, its apparent magnification changes and the focus plane shifts.

If the system operates near the depth-of-field limit, a tall object can become softer than a low one.

This matters especially when high speed has already forced short exposure and low signal. Defocus then becomes an additional reduction in effective contrast.

The OEM should qualify minimum and maximum object height at maximum belt speed rather than focus only on one reference plane.

Illumination Should Be Sized From Required Exposure, Not Visual Brightness

A SWIR light source that produces an excellent stationary image at 5 ms exposure may be completely inadequate for a production system requiring 80 µs.

The exposure ratio is:

5 ms / 0.08 ms = 62.5

Ignoring other factors, the high-speed configuration receives only about one sixty-second of the integration time.

The illumination, aperture and sensor response must make up that loss sufficiently to maintain useful signal.

This is why lighting should be specified only after the allowable exposure time has been calculated from conveyor velocity and object sampling.

Narrow-Band Inspection Can Reduce Data but May Require More Light

If spectral development shows that only a small number of wavelengths are necessary for classification, the production architecture may not need to process the complete spectral range at every object position.

Reducing spectral data can lower computational load and increase effective inspection throughput.

However, narrow-band optical filtering or wavelength-selective illumination can also reduce available photon flux.

The design therefore exchanges spectral data volume for optical signal.

The optimum solution is the minimum spectral information that preserves the required class separation at production speed.

Processing Latency Must Be Separated From Exposure Time

Exposure might last only 100 µs, but the classification pipeline may require several milliseconds.

The system must account for image transfer, preprocessing, spectral feature extraction, classifier execution, object tracking and output communication.

If the reject device is located sufficiently far downstream, several milliseconds may be entirely acceptable.

For example, at 2 m/s with the ejector 1 m downstream, nominal travel time is:

1 m ÷ 2 m/s = 0.5 seconds

That provides far more processing time than the exposure itself.

The machine architecture should therefore distinguish the optical time budget from the decision time budget.

Reject Timing Must Follow Conveyor Velocity

Once an object is classified, it must be rejected at the correct physical position.

For inspection-to-ejector distance (D) and object velocity (V):

Reject delay = D / V

If the camera is 800 mm upstream and the conveyor runs at 2.5 m/s:

0.8 / 2.5 = 0.32 seconds

If speed rises by 10% without updating the reject delay, the object reaches the reject station earlier.

High-speed sorting machines should therefore use reliable encoder-based position tracking or equivalent machine synchronization where speed changes materially.

A high-quality SWIR classification is wasted if the ejector acts on the next product.

Ejector Resolution Can Become the Throughput Bottleneck

The imaging system may identify two adjacent defective items perfectly, but the physical rejection system may be unable to actuate separately if they are too close.

Suppose an air ejector needs 10 ms between independent actions. At 3 m/s, objects separated by less than:

3000 mm/s × 0.010 s = 30 mm

can challenge independent rejection depending on actuator architecture.

The optical system should therefore not be specified independently of the reject mechanism.

Maximum practical inspection throughput is limited by the slowest critical subsystem.

Processing Only Relevant Regions Can Increase Throughput

Many high-speed scenes contain empty belt area.

Processing every pixel through a complex classifier wastes computing resources.

If object segmentation identifies only occupied regions, downstream spectral classification can be restricted to those areas.

Likewise, if only a subset of wavelengths is necessary, reducing unnecessary channels lowers data volume.

Current high-speed SWIR industrial systems explicitly use region-of-interest and spectral-selection approaches to increase acquisition or processing rates, illustrating the value of reducing data to what the application actually requires.

The same design principle applies regardless of the camera architecture: do not process information that cannot influence the pass/fail decision.

Conveyor Background Must Remain Spectrally Stable

A changing background can reduce classifier performance at high speed.

Wear, dust, product residue or moisture on the belt can alter its SWIR response. Mixed edge pixels then change over time even if product material remains identical.

This becomes particularly important with small objects because a significant fraction of their pixels can contain both object and belt information.

An OEM should therefore qualify the belt or chute background spectrally and include routine cleaning or reference correction in the machine-maintenance plan.

High-Speed Calibration Must Be Performed at Production Speed

A classifier calibrated only with stationary samples can hide motion-related problems.

Production qualification should include maximum conveyor speed, minimum exposure, final aperture, final illumination and realistic product spacing.

Compare the classification confidence of stationary and moving samples.

If confidence falls substantially at speed, inspect whether the cause is motion blur, lower signal-to-noise ratio, trigger variation, object rotation or insufficient processing time.

The production operating point—not the stationary laboratory image—should determine final acceptance.

Throughput Should Be Specified With Detection Performance

A claim such as “300 objects per minute” is incomplete.

A stronger specification is:

300 objects per minute while detecting the specified defect size at the required true-positive rate and maintaining the allowed false-reject rate.

If throughput is doubled but false rejects increase dramatically, the machine may become economically worse even though it is mechanically faster.

Inspection throughput should therefore always be linked to quality metrics.

Capacity Margin Prevents a Machine From Operating Permanently at Its Limit

An OEM should avoid designing every subsystem to operate at exactly 100% capacity during nominal production.

If required production speed is 2.5 m/s, qualification at a modestly higher speed can reveal whether enough optical and computational margin exists for normal fluctuations.

Similarly, if 100 parts/s is the commercial requirement, image processing and reject scheduling should ideally retain reserve capacity.

This margin helps accommodate future recipe complexity, normal conveyor variation and software updates.

Why Kyptec Automation® Is a Strong Platform for High-Speed SWIR Conveyor OEMs

The Kyptec Automation® SWIR Camera Lens collection provides five useful focal lengths—8.5 mm, 12.5 mm, 25 mm, 35 mm and 50 mm—within a consistent optical platform covering 900–1700 nm, F1.4, 2 MP, 2/3-inch format and C-Mount. The live collection currently contains these five dedicated models.

That range gives OEMs meaningful control over one of the central high-speed design variables: how many physical millimetres of conveyor are assigned to each sensor pixel. Short focal lengths can support broad multi-object scenes. Intermediate models can improve object sampling without forcing extremely narrow fields. Longer focal lengths can support fast-moving small components or controlled inspection channels.

The common F1.4 capability is also useful because short exposure frequently becomes necessary when product velocity is high. This does not eliminate the need for strong SWIR illumination, but it gives system designers valuable optical throughput margin.

Frequently Asked Questions About SWIR Camera Lenses for High-Speed Conveyor Inspection

1. How do I calculate the maximum exposure time for a moving conveyor?

First calculate object-side millimetres per pixel from the horizontal FOV divided by horizontal pixel count. Decide how much image motion is acceptable—often a fraction of a pixel for small-feature inspection—and convert that allowable physical movement into exposure time using exposure = allowable movement / conveyor speed. The final value should then be validated on real moving samples because classification can impose a stricter limit than visual sharpness alone.

2. How many pixels of motion blur are acceptable for SWIR inspection?

There is no universal limit. Tight dimensional or small-defect inspection may target less than one pixel and sometimes approximately half a pixel of movement during exposure. Large-object material classification may tolerate more. The correct limit is the point at which the smallest required material or defect feature still maintains adequate classification confidence.

3. Why does my SWIR image become noisy when I increase conveyor speed?

Higher speed normally forces shorter exposure to control blur. Shorter exposure collects fewer photons, reducing signal-to-noise ratio unless illumination or optical throughput is increased. This is a fundamental high-speed machine-vision trade-off: motion reduction and photon collection have to be balanced together.

4. Can opening the lens aperture compensate for shorter SWIR exposure?

It can help because a wider aperture passes more light to the sensor. The Kyptec Automation® SWIR Camera Lens family provides F1.4 maximum aperture across its current focal-length range. However, opening the aperture reduces depth of field, so products with substantial height variation should be tested before F1.4 is selected as the permanent production setting.

5. Does a wider FOV allow a faster conveyor?

Not automatically. A wider FOV increases object-space millimetres per pixel, which changes the motion-per-pixel relationship, but it also reduces the number of pixels across each product and defect. Conveyor speed can only increase if the resulting spatial sampling remains sufficient for the required inspection.

6. How do I calculate pixels per object on a SWIR conveyor system?

Calculate object-side sampling from FOV width ÷ horizontal sensor pixels, then divide the physical object width by that value. For example, a 400 mm FOV across 1600 pixels gives 0.25 mm/pixel. A 10 mm object therefore occupies approximately 40 pixels horizontally before optical blur and orientation are considered.

7. Why do small contaminants disappear when conveyor speed increases?

Motion blur spreads the contaminant signal across neighbouring pixels, while shorter exposures can also reduce signal-to-noise ratio. The combined effect can cause the contaminant spectrum to mix with the surrounding accepted material. The solution may require shorter exposure, stronger illumination, tighter FOV or more pixels on the contaminant.

8. Which Kyptec Automation® SWIR lens is suitable for a wide high-speed conveyor?

The Kyptec Automation® KL-1408 8.5 MM SWIR Camera Lens provides the widest focal-length geometry in the current range and is a strong candidate where maximum conveyor coverage is required. Its suitability should still be verified from the smallest object and required pixels per defect.

9. When should I choose the Kyptec Automation® KL-1410 instead of the 8.5 mm model?

The Kyptec Automation® KL-1410 12.5 MM SWIR Camera Lens can be a stronger choice when the 8.5 mm configuration provides more FOV than the process needs. Narrowing the field allocates more sensor pixels to each product while retaining relatively broad conveyor coverage, often improving segmentation and material sampling.

10. Is a 25 mm SWIR lens suitable for high-speed sorting?

Yes, when the sorting lane or object field is narrow enough. The Kyptec Automation® KL-1412 25 MM SWIR Camera Lens can provide stronger spatial representation of smaller products than a very wide lens. High speed and focal length are separate variables; the required FOV determines whether 25 mm is practical.

11. Can a 35 mm or 50 mm SWIR lens be used on a fast conveyor?

Yes. Longer focal lengths can be appropriate for small components moving quickly through a narrow controlled field or where greater working distance is required. The Kyptec Automation® KL-1414 35 MM SWIR Camera Lens and Kyptec Automation® KL-1416 50 MM SWIR Camera Lens provide these tighter geometries within the same 900–1700 nm platform.

12. Is frame rate the same as inspection throughput?

No. Frame rate describes image acquisition frequency, while inspection throughput describes how many objects can actually be classified correctly per unit time. One image can contain several products, and one product can require several acquisitions. Processing latency, object spacing and reject-system capacity also influence throughput.

13. How does object spacing affect high-speed SWIR sorting?

Insufficient spacing can cause overlapping segmentation regions and mixed spectral information. It also reduces the time between reject events. The minimum spacing should therefore be selected from both optical segmentation requirements and the response capability of the downstream rejection mechanism.

14. Should a SWIR conveyor camera run continuously or use external triggering?

Continuous acquisition can be appropriate for webs or densely populated streams. External triggering is often useful for discrete objects because it improves positional consistency and avoids unnecessary frames. The correct approach depends on product presentation and whether object timing can be measured reliably.

15. Why does increasing SWIR illumination sometimes work better than increasing exposure?

Increasing illumination can raise the number of photons collected without allowing the object to move farther during the exposure. Increasing exposure raises signal but directly increases motion blur. For fast-moving objects, additional illumination is therefore often the preferred method once the exposure limit has been established.

16. How should I calculate the camera-to-ejector delay on a sorting machine?

Divide the physical distance from inspection point to reject point by product velocity. If the distance is 0.75 m and belt speed is 2.5 m/s, nominal travel time is 0.30 seconds. If conveyor speed varies, position tracking or encoder-based synchronization is preferable to one fixed delay.

17. Can increasing image-processing speed compensate for motion blur?

No. Faster processing reduces decision latency but cannot restore spatial or spectral information lost during exposure. Motion blur must be controlled optically through exposure time, illumination, object sampling and motion geometry before software processing begins.

18. How should an OEM test maximum conveyor speed before releasing a SWIR inspection machine?

Test the machine with production-representative good and defective objects at progressively increasing speed while using the final illumination, aperture, FOV and processing pipeline. Track classification confidence, missed-defect rate, false rejection, trigger stability and reject accuracy. The maximum approved speed should retain acceptable margin rather than represent the point immediately before failure.

19. What information should be provided before selecting a SWIR lens for a high-speed conveyor?

Provide camera sensor dimensions and resolution, conveyor width, available working distance, maximum belt speed, smallest object, smallest defect or material region, object height variation, object spacing, required throughput and intended wavelength range. These values allow focal length, object sampling and maximum exposure to be solved together rather than selected independently.

20. Why is Kyptec Automation® a strong choice for high-speed SWIR conveyor systems?

Kyptec Automation® provides a focused SWIR Camera Lens portfolio with 8.5 mm, 12.5 mm, 25 mm, 35 mm and 50 mm focal lengths while maintaining a common 900–1700 nm, F1.4, 2 MP, 2/3-inch and C-Mount architecture. This gives OEMs practical flexibility to trade conveyor coverage against pixels per object while retaining the optical throughput needed for short-exposure SWIR inspection.

Conclusion

A SWIR camera lens for high-speed conveyor inspection should be selected as part of a complete motion-and-throughput budget rather than treated as an isolated optical component. Conveyor velocity determines how far the object moves during exposure. FOV determines object-side millimetres per pixel. Those two quantities establish how many pixels of blur occur at a given exposure time. Shortening exposure then reduces photon collection, which shifts the design pressure toward stronger illumination, efficient SWIR optics and an aperture that provides sufficient throughput without sacrificing the depth of field required by real products. Machine-vision guidance confirms this fundamental relationship between short exposure, motion blur and the increased lighting requirement for moving objects.

The strongest engineering workflow is therefore to start with the smallest production-relevant feature. Determine how many pixels it must occupy for reliable detection or classification. From that requirement, calculate the maximum FOV. Convert FOV into object-side pixel size, specify maximum allowable movement in pixels, and calculate the longest exposure permitted at maximum conveyor speed. Only then should illumination and aperture be optimized to achieve the necessary SWIR signal-to-noise ratio. After optical acquisition is solved, frame interval, processing latency, object spacing and reject timing can be engineered around the required objects-per-second throughput.

The Kyptec Automation® SWIR Camera Lens collection provides a strong platform for this process because its five focal lengths allow OEMs to move deliberately between broad conveyor coverage and higher object sampling without abandoning the dedicated 900–1700 nm optical family. The Kyptec Automation® KL-1408 can be evaluated where maximum conveyor width is important, the Kyptec Automation® KL-1410 offers a useful balance between coverage and pixels per object, the Kyptec Automation® KL-1412 can support smaller products or narrower high-speed lanes, and the longer-focal-length options can serve controlled fast-moving component inspection. The common F1.4 maximum aperture across the range is especially valuable when high belt speed forces short integration times.

For machine builders and industrial buyers, the most useful design principle is to calculate allowable motion before choosing exposure, and calculate object sampling before choosing focal length. A fast conveyor is not a problem if the optical system acquires enough clean information during the available time. Likewise, a high frame rate has little commercial value if objects are under-sampled, spectral contrast collapses at short exposure or the reject system cannot keep up. When exposure, motion blur, FOV, object sampling and throughput are engineered as one system, Kyptec Automation® SWIR Camera Lenses provide a technically strong optical foundation for reliable high-speed industrial inspection beyond visible light.