Back to search

FRIPRO-Fri prosjektstøtte

Measuring and Understanding Visual Appearance

Alternative title: Måling og forståelse av visuelt utseende

Awarded: NOK 12.3 mill.

Humans have the amazing capability to identify which material an object is made of, just by looking at it (for instance metal, plastic, rubber, wood, glass, or textile). In addition, also based only on its visual appearance, we can often evaluate key properties of the material, such as if it is fragile or robust, soft or hard, slippery or grippy, rotten or fresh, etc. This so-called material perception is an important function of the human visual system, and its evolution has enabled us to assess the environment surrounding us, and to survive in it. The visual appearance of a material is generally classified into four appearance attributes (colour, gloss, texture, and translucency) that interact with each other. This interaction is very complex and processed by the brain together with other information such as memory and viewing environment, to finally determine the perceived appearance of a surface or object. The concept of visual appearance is currently far from fully understood, neither from the metrological nor perceptual point of view. The goal of this project is to gain new knowledge of how human beings perceive the visual appearance of materials, objects, and scenes, and to develop new methodologies for measuring and communicating visual appearance and appearance-related material and object properties. The MUVApp project started in May 2016, and a kick-off meeting was organised in conjunction with the 4th CIE symposium on Colour and Visual Appearance in Prague in September 2016. The potential research directions and plans were discussed with the project members. Since the start of the project several workshops has been organized and research, training and dissemination activities conducted. More information about the project, including published papers can be found on Cristin (https://app.cristin.no/projects/show.jsf?id=536305). The research in MUVApp was organized in two main domains; Part A) Optical measurements and functions of appearance, and Part B) Visual processes of appearance understanding. In Part A, we have investigated the applicability of using an image-based technique to perform bidirectional reflectance measurements, and analytically estimate and represent material appearance. Homogeneous and flexible packaging materials with complex reflectance properties were used as test samples. Retroreflected light from the material surface was used to model material BRDF using analytical models which showed promising results for materials with gonio-chromatic and specular reflectance properties. We further investigated how contrast measures correlate with gloss perception. Since the acquisition of faithful colour data is an important component of appearance measurement, we developed an effective technique for absolute colorimetric camera characterisation, which can be used in a wide range of applications. A novel image-based technique was developed, for reverse engineering of woven fabrics at a yarn level, which can determine using a single image, the woven cloth structure and reflectance properties. The professional artistic production of computer-generated imagery (GCI) is typically a massively collaborative effort, in which appearance data and material models needs to be communicated, and where a broad range of modelling and rendering tools is involved. A lack of standards to exchange material parameters and data (between tools) requires the artists in digital 3D prototyping and design to manually match the appearance of materials to a reference image. To address the aforementioned issue, we developed a novel BRDF remapping technique, that automatically computes a mapping to match the appearance of a source material model. In Part B, a behavioural investigation of the assessment of visual appearance was conducted. The primary objective was to generate research hypotheses and outline future projects that eventually will lead to better understanding of appearance perception on a conceptual level. The study revealed significant role of shape, colorant concentration and surface coarseness in translucency and gloss perception. A study on the effect of blurring on translucency perception revealed the interesting trend that identical objects are perceived less translucent when observed in blurred images. Related to the goal of developing an international standard for ink opacity within ISO TC130, a psychophysical experiment was performed, where observers rated the opacity of white ink, with different ink film thickness, printed on different substrates. The proposed metrics showed good correlation with the visual assessments, and the metric of relative CIE Lightness provided the most linear result.

The project has helped establish the Colourlab as one of the world’s leading research groups in the interdisciplinary field of material appearance. Several new research projects have been established partly building on the knowledge and competence built in MUVApp (e.g. the INTPART project MANER, the MSCA ITN project ApPEARS, and the FRIPRO Young Researcher Talent project Spectraskin).

The Norwegian Colour and Visual Computing Laboratory at Gjøvik University College is an internationally leading research group in the field of colour imaging. Through the proposed project we will strengthen this position by broadening and deepening our area of research from colour to visual appearance. The concept of visual appearance, which includes colour but also other appearance properties like texture, gloss and translucency, is currently far from fully understood, neither from metrological nor perceptual points of view. The overall goal of our project is two-fold: to gain new knowledge of how human beings perceive the visual appearance of materials, objects, and scenes, and to develop new methodologies for measuring and communicating visual appearance and appearance-related properties. To reach this goal we propose a focused large scale research effort with innovative research approaches. Key aspects include the use of image domain metrics for characterizing appearance, the use of a combination of real world and virtual objects and surfaces for visual experiments, the study of interactions between appearance properties, and the development of new image-based methodologies for efficient yet precise measurement of physical properties that correlate with visual appearance such as the Bidirectional Surface Scattering Reflectance Distribution Function, the study of the correlation between physical measurements and perceived visual appearance. In addition to recruiting three new doctoral and two post-doctoral fellows, we include a significant international and inter-disciplinary collaboration with world-leading scholars in computer graphics, material science, and vision science. Combined with our leading scientific and technological research focus on imaging, including the recent rapid developments in the area of 3D printing, the gained control of the perception, measurement and eventually reproduction of appearance will have a very high societal impact.

Publications from Cristin

No publications found

No publications found

Funding scheme:

FRIPRO-Fri prosjektstøtte

Funding Sources