Machine vision motion blur: diagnose soft edges

Soft edges do not automatically mean you need a different camera. Motion during exposure, incorrect focus, limited depth of field, and image processing can produce similar-looking frames. The useful distinction is whether stationary detail is sharp and whether edge softness changes when exposure changes.

This procedure helps you isolate the cause using saved frames, a stationary reference, and controlled adjustments. Calculate the permitted exposure first, then test focus and lighting without changing several variables together. Replace hardware only when a measured requirement exceeds the current system’s capability.

1. Read the symptoms before changing settings

Inspect an original captured frame rather than a scaled video preview. Display it at native pixel size, disable viewer smoothing where possible, and compare several edge directions. Preview resizing and video compression can conceal fine detail or introduce their own artifacts.

Use this table to decide which test comes first. These patterns are clues, not proof; more than one cause can exist simultaneously.

First rule out a dirty optical surface, protective film, or an obstruction without disturbing the sensor assembly. Avoid aggressive cleaning or disassembly; follow the documented cleaning procedure for the component.

If the apparent blur exists only after resizing or geometric correction, review antialiasing during image downsampling and warping before adjusting optics.

2. Calculate the exposure budget from object motion

Frame rate is not exposure time. At 30 fps, frames begin approximately 33.3 ms apart, but an individual exposure can be much shorter. Raising frame rate alone does not guarantee a shorter exposure.

For approximately constant motion across a planar inspection field, estimate the object-space sampling and motion smear:

Use projected speed in the image plane, not an unrelated machine-axis speed. For diagonal movement, evaluate both image axes or use the corresponding vector magnitude. Rotating parts need a local surface-speed estimate at the feature being inspected.

Worked example

Suppose a captured image spans 240 mm across 1920 pixels. The sampling is:

Measure field width at the actual inspection plane. A different part height can change the sampling with a conventional lens.

At a projected speed of 500 mm/s and an exposure of 1 ms:

If the application allows no more than 0.5 pixel of motion smear:

The half-pixel limit is an example design target, not a universal acceptance criterion. A coarse presence check may tolerate more smear than a fine edge measurement.

Check sampling separately from blur. For example, allocating 4 pixels across a 0.5 mm feature over this field requires 240 / 0.5 × 4 = 1920 pixels. That allocation is only a starting point; detection reliability still depends on contrast, optics, noise, and the algorithm.

For a practical timing application, review sizing a strobe pulse for high-speed barcode verification. For measurement work, also consider how sampling affects edge-position accuracy.

3. Run a stationary-versus-moving test

Use the same target, optical geometry, and image path for every comparison. Choose a feature with useful contrast and retain its original pixel values. Automatic brightness changes can otherwise obscure the effect you are trying to measure.

1. Record the current configuration. Save exposure, gain, aperture, focus position, image dimensions, processing settings, and motion speed. Identify whether the saved image is raw acquisition data or processed output.
2. Capture the stationary reference. Stop the part using the machine’s approved operating procedure. Never reach into moving equipment or interfere with guarding to obtain a test image.
3. Set manual exposure and gain. Disable automatic controls for this diagnostic sequence. Keep sharpening, denoising, and compression unchanged, or disable them where the acquisition system permits.
4. Capture the moving target. Restore motion at the recorded speed without changing the optics. Compare the same feature and image location.
5. Shorten exposure in known ratios. Try a sequence such as 1 ms, 0.5 ms, and 0.25 ms. Increase controlled illumination as needed while keeping highlights below clipping.
6. Compare edge width and direction. Motion smear should decrease approximately with exposure when speed remains constant. Static optical softness will not follow the same proportional change.
7. Repeat with several frames. Retain at least 10 frames per condition as an initial diagnostic set. Variation between frames can reveal vibration, changing speed, or intermittent timing effects.

A single edge profile is often more informative than a whole-image sharpness score. Measure across an edge perpendicular to the expected smear direction, then compare an edge with a different orientation.

Use the same contrast range and processing when comparing profiles. An image-sharpness score can change with noise or texture rather than focus, so compare the same target region under controlled conditions.

4. Separate focus error from depth and optical limits

If the stationary target is soft, bracket focus before selecting another camera. Move focus through several small increments and save each result. A useful sweep passes through the sharpest position rather than stopping at the first acceptable preview.

Compare the center, corners, and actual inspection region independently. A sharp background does not demonstrate correct focus on the part. Likewise, a sharp center does not establish acceptable performance across the whole field.

If focus cannot reach the intended distance, check the mount, adapters, extension hardware, and documented working range. Review mount adapters, focus, and clearance rather than forcing the focus mechanism.

Next, test targets at the nearest and farthest expected part positions. If each becomes sharp at a different focus setting, depth of field is a likely limitation. Reduce height variation, alter the imaging geometry, or test a smaller aperture.

Stopping down involves a tradeoff. Changing from f/4 to f/8 reduces ideal image illumination to approximately one quarter, with other conditions unchanged. Recovering brightness through a longer exposure may restore motion blur. Very small apertures can also reduce fine-detail contrast through diffraction.

Test aperture changes against both a stationary target and the calculated exposure ceiling. Do not assume that the smallest opening provides the sharpest usable image. A sharp simulation preview is not sufficient evidence across a varying-height part; verify performance at the actual height extremes.

Browse machine vision lenses and use the lens finder only after establishing field coverage, working distance, mount, and sensor coverage. The LS-FFL-0602 is a C-mount lens specified for a 1" sensor format. Treat it as a catalog candidate, not a validated match for an unspecified inspection.

5. Recover light without giving back the exposure reduction

Once short exposure produces sharper moving edges, the remaining problem is often photon collection. Gain increases the recorded signal level but does not collect more photons or recover detail already smeared during exposure.

In the worked example, reducing exposure from 1 ms to 0.125 ms cuts integration time by a factor of 8. At unchanged aperture and sensitivity, maintaining a comparable signal requires approximately 8 times the effective irradiance at the target, assuming linear response and no clipping.

Start with lighting geometry. Move illumination so the inspection feature has useful contrast, avoid direct reflections into the lens, and check coverage across the full region. Bright average intensity is not helpful when the required edge disappears in glare.

Bar lights allow directional lighting arrangements, but their suitability depends on the surface and inspection task. The LS-BAR-1047 is a standard LED bar light. Its catalog identification alone does not establish whether it can meet your measured short-exposure requirement.

Use the lighting finder to evaluate appropriate geometry and compare test images before committing to an arrangement. For illumination across the inspection area, review balancing lighting coverage and brightness.

Strobing can limit the effective illumination period, provided ambient light contributes negligibly and the pulse overlaps the intended exposure. Electrical pulse duration is not automatically the useful optical duration. Follow the procedure for measuring LED strobe optical rise time, and operate lights and controllers only within their documented ratings.

6. Decide whether the camera is actually limiting you

A camera change becomes justified when a required capability is missing: usable exposure control, suitable shutter behavior, adequate sampling, synchronization, or sufficient signal quality at the permitted exposure.

Global shutter removes row-to-row exposure timing differences, but it does not eliminate motion during the exposure itself. Rolling shutter can introduce geometric distortion in addition to ordinary smear. Reducing exposure may improve smear while leaving some readout-related distortion visible.

The LS-free camera selection guide is not required here is not a substitute for measured requirements. Instead, compare area scan cameras against the exposure budget and motion behavior you have established.

The MV-CU020-90GC uses a global shutter and a GigE interface. Confirm the complete timing, control, sensitivity, and optical requirements before treating it as a replacement candidate.

Do not assume an arbitrary strobe will expose every row uniformly on a rolling-shutter camera. Confirm the documented exposure-overlap and synchronization modes. A shutter change is the wrong remedy when the stationary image remains out of focus.

Likewise, additional pixels do not automatically fix blur. With unchanged field width and exposure, finer sampling spreads the same physical movement across more pixels. Review exposure, gain, and image-processing settings when the remaining limitation is noise or clipping rather than motion.

7. Verify the fix before ordering or releasing the station

Validate the chosen settings at the highest relevant projected speed, expected part-height extremes, and least favorable target contrast. Include normal production variation rather than judging only a carefully positioned reference part.

Use this final checklist:

• Original frames: confirm the improvement exists before display scaling, compression, or downstream enhancement.
• Exposure readback: verify the applied setting, not just the value requested by software.
• Motion budget: recalculate smear using measured field width and maximum relevant speed.
• Focus repeatability: confirm stationary sharpness after normal adjustment and restart procedures.
• Height range: inspect both nearest and farthest part positions without refocusing.
• Lighting margin: check that useful edges remain visible without clipping or excessive gain.
• Timing behavior: inspect consecutive frames for intermittent smear, distortion, or illumination bands.
• Inspection performance: verify measurement error or detection decisions, not only visual attractiveness.

Record the passing configuration and retain representative images for future comparisons. If the test identifies an optical or lighting limitation, start the camera configuration process with the measured requirements, or contact applications support with stationary and moving frames, field dimensions, speed, exposure, and the required inspection result.