Measuring and modelling Surface Appearance for the surface topography control – NAPS

Mastering the appearance function of manufactured surfaces—that is, the objectification and management of subjective information through industrial processes—is a scientific and technical challenge. The visual perception of surfaces is a complex and unconscious process, influenced by our knowledge, experiences, objectives, and emotions. However, appearance also involves sensory notions related to metrological and physical aspects, requiring a deep understanding of multi-physical and multi-scale interactions.

A first, geometric approach, consists of modeling visual behavior based on the measurement of three-dimensional surface information at the microgeometric scale, using physical or geometric models. However, this approach requires very fine spatial and vertical measurement resolution, which results in data measurement, processing, and analysis times that are often prohibitive in an industrial context.

A second, photometric approach, models visual behavior based on the measurement of physical appearance attributes related to light re-emission. The most comprehensive technique is the determination of the bidirectional reflectance distribution function (BRDF), but it is time-consuming and unsuitable for industrial inspection. Methods based on restrictions on the global BRDF model have been initiated, but they are associated with an assumption of homogeneity of behavior that is inappropriate for inspection.

The NAPS project proposes a local method, based on the Reflectance Transformation Imaging (RTI) technique, which reduces the global BRDF model to the angular and spectral components of reflectance. The challenges of this approach include measuring appearance attributes at the scale of manufactured surface finishes, modeling appearance information, characterizing and qualifying the appearance of a surface finish, and establishing correlations between process and functionality. The NAPS project was structured around these challenges, with a cross-cutting challenge related to the selection and determination of relevant scales, aimed at providing answers both instrumentally and methodologically.

As part of the project, the chosen axis was to focus on the angular reflectance component, which is essential for surface inspection and in particular the evaluation of local appearance anomalies. The technique implemented is called Reflectance Transformation Imaging (RTI). It was shown how this technique can be implemented to metrologically digitize information and how it can be used to model the local angular reflectance of surfaces, and thus extract relevant objective information for efficient management of the appearance of manufactured surface conditions. An innovative appearance digitization system based on the RTI technique at the surface condition scale was developed. This device integrates all the methods developed during the NAPS project, and in particular the coupling of RTI and High Dynamic Range imaging techniques. The information thus digitized makes it possible to discriminate local behaviors in terms of appearance on the inspected surfaces, particularly at the location of anomalies (criticality assessment). An industrial development process was initiated following this project with industrial partners from the national network.

A collaboration has been formalized with a French manufacturer of surface finish metrology instruments (Altimet) with the aim of industrializing the manufacturing of the device and marketing it (an entity already sold, currently being assembled). The prospects for industrial development of this work are therefore quite tangible and promising.

The scientific aspects related to functionalization, that is, the implementation of statistical relevance analysis methods to link measured data to local surface appearance characteristics, have also been implemented with significant results. The positive results obtained in the initial approach open up very positive prospects for continuing this work, including the opportunity to implement a deep learning approach to link the data obtained to surface behavioral parameters.

Nurit, M.; Le Goïc, G.; Lewis, D.; Castro, Y.; Zendagui, A.; et al. HD-RTI: an adaptive multi-light imaging approach for the quality assessment of manufactured surfaces. Computers in Industry. 2021, 132, 103500.

Castro, Y.; Nurit, M.; Pitard, G.; Zendagui, A.; Le Goïc, G.; et al. Calibration of spatial distribution of light sources in reflectance transformation imaging based on adaptive local density estimation. Journal of Electronic Imaging. 2020, 29(04), 1.

Zendagui, A.; Thomas, J.-B.; Le Goïc, G.; et al. Quality assessment of reconstruction and relighting from RTI images: application to manufactured surfaces. 2019 15th International Conference on Signal-Image Technology & Internet-Based Systems (SITIS), Sorrento, Italy. 2019, 746-753.

Nurit, M.; et al. Improved visual saliency estimation on manufactured surfaces using high-dynamic reflectance transformation imaging. Fifteenth International Conference on Quality Control by Artificial Vision, May 2021, Tokushima, Japan. 2021, 51.

The measurement and modelling of surface visual appearance is a challenge and an important issue at both scientific level and in their industrial applications, as quality inspection, and controlling the surface manufacturing and finishing processes :
(i) how to measure visual appearance indicators, how to qualify a surface appearance from objective and robust indicators
(ii) how to determine functional correlations between the surface appearance and the surface micro-geometry
The NAPS project propose to answer these two issues, at both hardware and methodology levels, by the design of an innovative device to measure surface appearance (restricted to the angular and spectral component of the reflectance), and by the implementation of methods that allow to reconstruct the angular reflectance efficiently. The proposed methods also aim to make easier the determination of functional correlations between the surface appearance and its geometry.