Overview
ISO 16610-71:2014 - Geometrical product specifications (GPS) - Filtration - Part 71: Robust areal filters: Gaussian regression filters - defines the metrological characteristics and equations for a robust areal Gaussian regression filter used to evaluate measured surfaces that may contain spike discontinuities, deep valleys or high peaks. The standard describes both planar and cylindrical variants and specifies how to separate large-scale lateral components (form and waviness) from short-scale lateral components (texture).
Key topics and technical requirements
- Robust areal regression filter: A weighted M‑estimator based on local complete polynomial modelling (default polynomial degree p = 2) used for areal (2D) surface filtration.
- Areal Gaussian weighting function: Spatially varying Gaussian weights used in regression; the filter combines spatial position and measured height values.
- Biweight influence function: The filter uses an asymmetric biweight function to compute weights (δ) that reduce the influence of outliers (spikes, deep valleys).
- Scale parameter (c): Default c is derived from the Median Absolute Deviation (MAD) of residuals (relation to inverse error function provided in the standard), giving robust sensitivity to amplitude distribution.
- Filter equations and implementation notes: Matrix-based regression equations, iterative weight updates (start with δ = 1; update via biweight), and special-case p = 0 to increase robustness at boundaries.
- Planar and cylindrical forms: Separate formulations for nominally planar and cylindrical surfaces (including radial/circumferential coordinates for cylinders).
- Transmission characteristic: Because weights depend on surface values and location, no single transmission characteristic is provided.
- Supplementary material: Annexes include mathematical definitions, examples, and relationship to the GPS matrix model; a Nesting Index and filter designation rules are specified.
Applications
- Surface texture and topography analysis where measurements include spikes, pores, peaks or boundary effects.
- Pre‑processing of areal measurement data from tactile (stylus) or optical (confocal, white-light) instruments to separate form/waviness from texture.
- Quality control and specification of functional surfaces (bearing surfaces, seals, additive‑manufactured parts) where robust filtering avoids distortion by extreme local features.
- Implementation in metrology software and instrumentation firmware for consistent, GPS-compliant areal filtration.
Who uses this standard
- Metrologists and surface texture analysts
- Inspection and quality engineers
- Measurement instrument manufacturers and software developers
- Standards committees and test laboratories
Related standards
Keywords: ISO 16610-71:2014, robust areal filters, Gaussian regression filters, filtration, surface metrology, biweight function, MAD, areal filtering, GPS.