Correlation matrices

A diagnostic tool for looking under the hood of a single vector — the raw correlation surface that produced it, for every pass.

Open via Plot → Correlation matrices.

PIVlab Display correlation matrices panel — The **Display correlation matrices** panel.

Click Retrieve correlation matrices to compute them for the currently displayed frame.
Click on any vector in the plot to display the correlation matrices of all passes for that specific vector.

Pop-up showing four correlation matrices for one vector, one per pass, each with a single bright peak; the interrogation window shrinks from 128x128 down to 20x20 across the passes — The correlation matrices for a single vector, one per pass. The interrogation window shrinks each pass (128×128 → 20×20 here), and each surface has a clear, single peak — the sign of a reliable vector.

What the surface looks like tells you why a vector turned out the way it did:

Peak shape	Meaning
Sharp, single, dominant peak	A reliable match — this is what you want to see.
Multiple competing peaks of similar height	An ambiguous match — the correlation can't clearly tell which displacement is correct.
Broad, flat peak	Poor accuracy — the location is only loosely determined.
Noisy surface with no clear peak	Too few particles or poor image quality in that interrogation area.

When this is useful This explains why a particular vector looks wrong or got flagged during validation — useful for understanding, not just seeing, a bad result.