Knowledge base

Image measurement, explained properly.

Seven guides on the parts of image-based metrology that actually decide whether a number is trustworthy: what a calibration claims, where accuracy is lost, which distance a drawing means, and what an auditor will ask you for.

Written from the software outwards.

Every guide here is grounded in how Measuret actually works — the same calibration transform, the same tolerance states, the same report columns. Where something is a general metrology principle rather than a product feature, it says so.

Calibration

What "mm per pixel" really means

A calibration is a claim about your whole optical chain at one moment. How to derive it honestly, and the five ways the figure quietly goes wrong.

Calibration

Why non-square pixels break naive measurement

When X and Y do not scale equally, a circle is an ellipse in pixels and every diagonal is mis-measured. How to detect it in ten minutes.

Quality

Traceability: what an auditor looks for

The chain from a national standard to your reading, the link that image measurement adds, and the six questions auditors actually ask.

Optics

Magnification vs resolution vs accuracy

Three different things, routinely confused. Why zooming in further often measures worse, not better.

Reporting

Why your report needs a scale bar

"200×" stops being true the moment an image is resized, cropped or pasted. A scale bar is made of the same pixels as the part.

Method

Shortest distance vs a fixed direction

Two operators, one image, two different answers — and neither made a mistake. Which one the drawing actually asked for.

Tolerance

Pass, Fail, and why Indeterminate exists

The dangerous verdict is not Fail. It is a Pass that was never earned, because the column had nowhere else to put an unmeasured row.

Common questions

Short answers.

Longer ones are in the guides above. If your question is not here, ask an engineer.

How accurate is image measurement?

There is no single figure, and any supplier who gives you one without asking about your part, optics and lighting is guessing. Accuracy depends on the calibration, the pixel scale, how repeatably an edge can be located under your lighting, lens distortion across the field, and whether the part sits in the plane you calibrated in. The honest approach is to work out what your tolerance demands and then specify optics that comfortably exceed it.

Do I need a microscope, or will a camera do?

It depends entirely on feature size and tolerance. What matters is the resulting pixel scale and edge quality, not the label on the instrument. A good telecentric lens on a machine-vision camera will out-measure a poorly lit microscope for many production parts; a microscope is necessary when the features are genuinely small.

How often should I recalibrate?

On a schedule your quality system sets, and whenever anything optical changes: objective, zoom, working distance, camera, or a fixture that alters the height the part sits at. The second list is the one that catches people out, because none of those events look like a calibration event at the time.

Can Measuret measure things it cannot see?

No, and neither can anything else optical. Image measurement measures a silhouette or a visible boundary. Depth, hidden features, internal geometry and anything obscured by the part itself are outside what a single image contains. If a drawing calls those out, you need a different instrument for those dimensions.

Does Measuret do weld inspection or image stitching?

No. Measuret is deliberately scoped to geometric measurement and excludes weld inspection, image stitching, extended depth of field and morphological image processing. Those belong to other Multitek products — ask us and we will point you at the right one.

What units does it work in?

Micrometres, millimetres, centimetres, metres and inches. The unit is part of the calibration, and it is a column on the report rather than an assumption.

Knowledge

Something missing?

If there is a measurement question you keep having to explain to new starters, tell us — it is probably worth a guide, and we would rather write the one people actually need.