Missing Point Interpolation Algorithms for Areal Texture Measurement ISO 25178
ISO 25178 parameters require raw height matrices capped at three percent missing points to prevent severe hybrid parameter smoothing and false tooling acceptance.

Void
Optical topography acquisitions on steel tool inserts frequently drop height coordinates near steep side-walls, re-entrant angles, or unpolished laser-ablated textures. Specular reflection fails when slope angles exceed the numerical aperture of objective lenses. Photodetectors record zero reflected intensity, generating unmeasured pixels inside spatial arrays.
Micro-structured surfaces cut by electrical discharge machining or picosecond laser ablation present severe light scattering. High surface roughness produces dropouts precisely where height transitions are most sharp.

Non-Measured Points in Optical Topography Acquisition
Surface slope dictates specular reflection. Light fails to return. When measuring micro-textured injection tooling or photo-etched grain patterns, coherence scanning interferometers and focus variation microscopes hit physical collection limits.
Non-measured points appear as empty cells within the rectangular elevation matrix. Unfilled matrix cells break spatial frequency calculations, fast Fourier transforms, and derivative-based hybrid parameters under ISO 25178-2.
Zeroes replace missing data. Standard numerical software defaults often convert unmeasured coordinates into zero height values. This substitution creates extreme artificial gradients between valid topography features and missing pixels.
Downstream spatial filters treat these step changes as real structural features, generating ringing artifacts and severe parameter distortion. Topography evaluation demands controlled handling of missing coordinates before extracting standard areal parameters.
- Sensor threshold gating separates valid height returns from detector background noise.
- Spatial mapping flags unmeasured coordinate locations within the spatial matrix as null values rather than numerical zero.
- Neighborhood density evaluation calculates the ratio of valid surrounding pixels within a designated boundary kernel.
- Algorithm selection routes isolated missing coordinates to localized estimators while assigning wide continuous clusters to global boundary conditions.
- Height value substitution updates null grid locations and assigns process flags to track modified height points.
Coherence scanning interferometers running on 50x objectives lose directional reflection when local surface gradients exceed 27.5 degrees relative to the optical axis.
Unfilled data patches vary in size depending on surface finish and instrument technology. Coherence scanning interferometers struggle with high local slope gradients, whereas focus variation optics lose signal in flat, mirror-polished regions lacking optical contrast. Micro-injection moulding inserts combining mirror-polished land areas with steep micro-fluidic channels exhibit both failure modes simultaneously.
Data drops out here.
Executing texture audits on height matrices containing unflagged missing points yields false acceptance of underscratched or over-ablated tooling surfaces, passing defective mould geometry into high-volume manufacturing.

Grid
Mathematical reconstruction of missing coordinate points relies on spatial neighborhood relationships within discrete height arrays. Interpolation algorithms estimate unrecorded elevations by evaluating surrounding valid coordinates. The choice of interpolation scheme alters the spatial frequency spectrum and local curvature of the reconstructed surface topography.

Spatial Reconstruction Mechanics across Non-Continuous Surfaces
Interpolation estimators construct localized height estimates by solving algebraic, differential, or distance-weighted polynomial functions. Nearest neighbor methods assign the value of the closest valid coordinate to the missing location. Nearest neighbor calculations preserve original measured height values without introducing out-of-range numerical extremes, but create sharp planar steps.
The profile flattens instantly.
Bilinear interpolation calculates unknown elevations using linear distance weighting across four surrounding valid pixels. Linear schemes smooth local step transitions but attenuate high-frequency surface roughness. Higher-order methods, including bi-cubic spline and radial basis function interpolation, evaluate broader spatial neighborhoods to produce continuous first and second spatial derivatives across filled gaps.
| Interpolation Method | Mathematical Basis | High-Frequency Loss | Computation Overhead | Recommended NMP Limit |
|---|---|---|---|---|
| Nearest Neighbor | Zero-order hold | Zero attenuation | Minimal | 1.0% |
| Bilinear | First-order polynomial | Moderate smoothing | Low | 3.0% |
| Bi-Cubic Spline | Third-order piecewise polynomial | Low smoothing | Moderate | 5.0% |
| Radial Basis Function | Distance-based kernel functions | High smoothing | High | 2.0% |
| Planar Regression | Local least-squares fit | Severe smoothing | Low | 1.5% |
| Performance evaluated on VDI 3400 texture Grade 24 tool steel matrices under ISO 25178-2 protocol parameters. | ||||
Bicubic splines overshoot narrow peaks. When missing points cluster near high-aspect-ratio ridges or sharp laser-etched edges, higher-order polynomials generate numerical ringing. Reconstructed height values can exceed physical peak heights or fall below valley bottoms.
Modal decomposition and morphological filling operators stabilize reconstruction by enforcing geometric shape constraints derived from valid surrounding topography.
- Peak clipping under bilinear estimation flattens local asperities and alters calculated peak density values across micro-structured tool cavities.
- Bicubic overshoot near step edges introduces artificial oscillations into flat plateaus surrounding deep textured valleys.
- Nearest neighbor tiling creates artificial step discontinuities that corrupt spatial autocorrelation length computations.
- Radial basis function smoothing dampens high-frequency tool marks and disguises diamond tool chatter patterns.
ISO 25178-600 requires that any mathematical modification of measured height matrices be declared alongside the reported texture parameters.
Tooling suppliers maintain that automated software smoothing removes optical noise without altering underlying tool geometry.

Distortion
Parameter calculation routines specified under ISO 25178-2 exhibit divergent sensitivity profiles when processed over filled height matrices. Height parameters based on statistical amplitude distributions maintain stability under low missing point ratios. Hybrid parameters dependent on surface slopes and areal integration suffer immediate, severe skew when missing points are reconstructed.

Parametric Drift in ISO 25178 Field Metrics
Calculated height metrics react differently to substituted point values depending on whether spatial derivatives or absolute heights govern the parameter definition. Arithmetic mean height, designated Sa, averages absolute elevation deviations across the evaluation area. Sa exhibits minimal drift under low percentages of interpolated data because missing points contribute proportionally to total sample area.
Root mean square gradient, designated Sdq, computes spatial derivatives between adjacent pixel locations across X and Y axes. Gradient calculations amplify local noise. When an interpolation algorithm smooths out a sharp pit transition or creates a polynomial ripple, Sdq deviates significantly from true surface values.
Developed interfacial area ratio, designated Sdr, combines spatial gradients to express additional surface area created by texture compared to a flat plane. Sdr metrics collapse under over-smoothed linear or spline estimations.
| Parameter | Parameter Type | 2% NMP Error | 5% NMP Error | 15% NMP Error | Dominant Drift Direction |
|---|---|---|---|---|---|
| Sa | Height (Amplitude) | +0.1% | -0.8% | -3.4% | Negative shift |
| Sq | Height (Root Mean Square) | +0.2% | -1.1% | -4.2% | Negative shift |
| Sz | Height (Extreme Peak-Valley) | -1.5% | -6.2% | -18.9% | Severe contraction |
| Sdq | Hybrid (Spatial Gradient) | -3.8% | -12.4% | -38.1% | Severe attenuation |
| Sdr | Hybrid (Interfacial Area) | -7.1% | -21.5% | -56.3% | Severe attenuation |
| Vmc | Functional (Core Volume) | +0.4% | +1.9% | +8.5% | Positive shift |
Errors propagate into functional volume ratios. Functional void volume parameters, including peak material volume Vmp and dale void volume Vvc, rely on material ratio curves constructed by sorting height matrices from highest peak to lowest valley. Replacing unmeasured valley points with smoothed interpolation values artificially reduces calculated oil retention volumes on engine cylinder liners and textured seal faces.
- Arithmetic mean height metrics resist minor point substitutions due to global integration averaging across the full evaluation area.
- Root mean square gradient metrics scale rapidly with spatial position jumps introduced by coarse interpolation schemes.
- Developed interfacial area ratios collapse toward baseline surface areas when smooth splines bridge deep micro-grooves.
- Functional volume parameters shift systematically as missing valley points receive overestimated fill values.
Field parameters driven by spatial derivatives experience far greater calculation instability from interpolated points than parameters derived purely from height distribution histograms.
Clause 6.4 of ISO 25178-2 dictates that missing point percentages exceeding five percent invalidate standard parameter reporting, forcing manual verification before tool sign-off.

Threshold
Filtering workflows and data pre-processing steps defined in ISO 25178-3 dictate the exact order of operators applied to raw sensor matrices. Interpolation timing relative to bandpass filtering governs whether local reconstruction errors stay localized or spread across the evaluation field. Applying an S-filter or L-filter over missing coordinate cells distributes numerical null values into surrounding valid pixels.

Where Does Interpolation Distort Functional Volume Parameters?
Core fluid retention metrics derived from material ratio curves show extreme sensitivity when missing data concentrates near deep surface valleys. Optical sensors lose signal in deep grooves where returned light cannot clear sidewall geometry. Standard software interpolation bridges these voids, effectively raising the valley floor.
Calculated dale void volumes drop, distorting fluid capacity forecasts for moulded sealing surfaces.
Pre-processing protocols mandate executing missing point replacement prior to surface filtering. The spatial operator order establishes that raw data gating occurs first, followed by isolated point interpolation, nested S-filtering, form removal, and secondary L-filtering. Reversing this sequence allows spatial filter convolution kernels to incorporate unmeasured values, distorting the target surface geometry across several filter cut-off lengths.
- Missing point threshold setting caps allowed unmeasured coordinates at three percent of total dataset points prior to parameter computation.
- Cluster size bounding isolates single missing coordinates for linear fill while masking multi-pixel gaps against artificial surface creation.
- Operator sequence enforcement executes non-measured point filling prior to S-filter application to prevent edge artifact propagation.
- Dual dossier archiving saves raw un-interpolated height arrays alongside final parameter extraction files for auditing.
Optical metrology software often applies default interpolation routines automatically during file loading unless operator overrides are explicitly saved in the measurement template.
The height map corrupts. Uncontrolled software defaults disguise severe data loss under aesthetically smooth 3D surface renders. Quality engineers must inspect raw, unfilled point density maps before signing off on metrology reports generated for micro-structured tool cavities.
Metrologists continue to debate whether adaptive neural-network surface reconstruction presents superior physical fidelity compared to deterministic bi-cubic spline operations on complex EDM textures.

Ledger
Commercial agreements governing injection tooling and micro-structured inserts require unambiguous definitions of metrology protocols to prevent costly sign-off disputes. A toolmaker presenting surface roughness data based on aggressive 15% interpolation can pass an insufficiently textured insert that later fails to release plastic parts during moulding. Molders require transparent data handling protocols attached to tooling invoices.

Tooling Sign-Off Audits and Metrology Verification
Quality agreements written for high-precision plastic optics stipulate exact measurement conditions, sensor objective choices, and data-cleaning allowances. Injection-moulded light guides and diffusers depend on pristine tool micro-geometry to maintain optical performance. Interpolation algorithms that smooth diamond-turned optical tool inserts conceal micro-chatter, leading to light leakage in final moulded parts.
| Quality Stage | Surface Metric Focus | Max Allowed NMP | Mandated Data Handling |
|---|---|---|---|
| T1 Sample Inspection | Sa, Sz, Sdq | 1.0% | Bilinear fill only; raw matrix archived |
| Tool Steel Audit | Sdq, Sdr, Vmc | 2.0% | Bi-cubic spline; cluster cap 5×5 pixels |
| Textured Grain Review | Sa, Sal, Str | 3.0% | Modal decomposition; report raw dropouts |
| High-Gloss Mold Insert | Sq, Sp, Sv | 0.5% | Zero interpolation permitted |
Tooling procurement contracts set explicit numerical caps on missing point density and define acceptable interpolation routines. Tooling sign-off requires raw height matrices. Supplier inspection reports omitting raw point density statistics fail initial quality entrance audits.
Quality managers enforce procurement agreements that mandate submitting native optical file formats containing full acquisition metadata alongside generated PDF reports.
Precision moulding operations maintain dimensional stability only when original tool steel topographies are recorded without mathematical distortion.




