Our work  /  Case study

Automated sieve measurement and calibration in ZEN Blue

Procertus, sieve certification specialist, Germany
Field engineers certifying production sieves on-site at cement plants

Field
Cement production · Materials testing
Instrument
ZEISS manual microscope · ZEN Blue
Stack
.NET · ZEN Blue API · OpenCV · CSV / JSON
Context

Automated sieve calibration by microscope

Procertus performs on-site sieve measurement and calibration for cement production facilities. Using a ZEISS microscope, the workflow measures aperture size, wire thickness and sieve uniformity at micrometer scale before the customer generates the applicable report and certificate.

The work is done on-site. The engineer assembles the microscope at each customer, which invalidates any lab calibration, so a fresh calibration runs once per visit before measurement. Calibration uses a traceable plate: reference squares of known size are imaged and measured to establish an exact pixel-to-micron ratio, which is then applied to every measurement in that session.

Problem

Why manual aperture and wire measurement took two hours

The process ran on Axio Vision. Operators aligned manually before the software measured and recorded the calibration data for each run. Calibration alone took ten steps with four measurements each, around 20 to 30 minutes. A sieve carries 100 to 200 holes per image and is typically imaged twice, so 200 to 400 holes had to be identified and measured. A full measuring and certification pass came to roughly two hours.

Identifying holes and wires by hand could introduce run-to-run variation, and the scale set during manual calibration carried into every measurement that followed. Calibration data, images, and results were also stored separately, which made a certificate hard to trace back to its evidence.

Constraints

Working around a non-motorized, on-site setup

01
Non-motorized stage

The plate and the sieve are moved by hand, so calibration cannot run automatically.

02
On-site calibration

A fresh calibration is required at each site visit, before any measurement.

03
Customer reporting

By design, the certificate stays in the Procertus Excel workbook, which holds the formulas.

Solution

Automatic sieve aperture and wire measurement

SmartLabs delivered a guided workflow as a .NET extension inside ZEISS ZEN Blue, replacing the Axio Vision application. It runs entirely within the customer's installation, with no external process and no network dependency in the field.

Calibration

Calibration is optional per session: reuse the last one, or run a new one. Because the stage is not motorized, the operator brings each plate pattern into view in a set order while the software detects the reference squares and fixes the scale. Supported plate sizes run from 20 to 4000 µm, and a dusty sample that blocks detection prompts a redo.

Measurement

The operator selects the sieve by aperture and sets two parameters, image count and Z-stack. Detection runs on classical OpenCV (thresholding, then contour detection, then geometric feature analysis) and returns hole diameter and wire thickness deterministically, so the same image yields the same numbers on every run. Warped sieves are captured as a Z-stack, reading thickness at the correct depth rather than from a single plane.

Handoff to the certificate

Measurements of the sieve holes, wire diameters, calibration, and the smallest and biggest hole are written together to structured project folders and exported as CSV and JSON into a dedicated Excel workbook connected to the solution. The operator then transfers this data into the Procertus workbook, which holds the certificate formulas. That CSV/JSON format is the only handoff point we created, so the workflow and the customer's workbook can each change without affecting the other.

A Procertus engineer at a ZEISS microscope on-site, with the sieve mesh and its detected apertures displayed in the workflow on a laptop beside the instrument
The workflow in use on-site: the sieve under the ZEISS microscope, with the mesh and its detected apertures on screen beside it.
Technical stack

Technical stack: OpenCV in ZEN Blue

Platform
ZEISS ZEN Blue, automation via a .NET extension through the ZEN Blue API
Image processing
Classical OpenCV: thresholding, contour detection, geometric feature analysis
Exports
CSV and JSON into structured project folders, feeding the customer Excel workbook
Hardware
ZEISS manual microscope, non-motorized XY stage, PlanApo Z 0.5x with motorised focus
Delivered
3 to 4 months
Result

Results: 7–10 minutes per sieve

7–10 min
To measure and certify one sieve, previously around two hours
~0.5%
Average measurement error, 1% maximum
200–400
Holes measured per sieve, detected rather than identified by hand
BeforeAfter
Time per sieve2 h, manual with Axio Vision7 to 10 min, guided and automated
Calibration20 to 30 min by handDetected and scaled per session
MeasurementAxio Vision with manual identificationDeterministic OpenCV detection
Measurement errorVariable~0.5% average, 1% maximum
TraceabilityCalibration, images, results stored separatelyLinked in folders, CSV and JSON export

No change was made to the microscope or to the customer's reporting workbook.

Client testimonial
"A reliable partner who delivers not only outstanding quality but also proactively contributes to improving our processes. Thanks to their expertise, we have significantly reduced the time required for sieve calibration, resulting in greater operational efficiency while maintaining the highest quality standards."
Koen Derboven Metrology Service Manager, Procertus
Applicability

Other uses for automated microscope metrology

Nothing here is specific to sieves. The same approach fits any case where precision measurements are taken in the field, on a non-motorized instrument calibrated fresh on-site, and the results feed a certificate.

ContextThe problem we could help with
Filter mesh and textile inspectionRepeating aperture structures checked against a tolerance, the same hole and wire geometry
Tissue scaffolds, porous membranesPore diameter and strut thickness measured against a spec, the same repeating geometry as a sieve mesh
On-site metrology, other industriesInstrument assembled at the customer, prior calibration invalid, fresh calibration per visit
Non-motorized ZEN Blue setupsSame scripting surface (.NET, ZEN Blue API), manual stage, no native guided calibration and measurement
Calibrated dimensional QCAny chain where a traceable reference sets the scale applied across a run, with the output feeding a certificate
Warped or non-planar samplesA Z-stack is needed to recover true dimensions where one focal plane misreads

Working on something similar?

If measurement, calibration, and reporting are still manual steps in your process, describe the workflow and we will tell you what is feasible.