Mathematical Models for Perceived Roughness of Three-Dimensional Surface Textures

Research output: Book/ReportCommissioned report


This thesis is concerned with the visual perception of glossy rough surfaces, specifically those characterised by 1 / f ß noise.

Computer graphics were used to model these natural looking surfaces, which were generated and animated to provide realistic stimuli for observers. Different methods were employed to investigate the effects of varying surface roughness and reflection model parameters on perceived gloss.

We first investigated how the perceived gloss of a matte Lambertian surface varies with RMS roughness. Then we estimated the perceived gloss of moderate RMS height surfaces rendered using a gloss reflection model. We found that adjusting parameters of the gloss reflection model on the moderate RMS height surfaces produces similar levels of gloss to the high RMS height Lambertian surfaces.

More realistic stimuli were modelled using improvements in the reflection model, rendering technique, illumination and viewing conditions. In contrast with previous research, a non-monotonic relationship was found between perceived gloss and mesoscale roughness when microscale parameters were held constant. Finally, the joint effect of variations in mesoscale roughness (surface geometry) and microscale roughness (reflection model) on perceived gloss was investigated and tested against conjoint measurement models. It was concluded that perceived gloss of rough surfaces is significantly affected by surface roughness in both mesoscale and microscale and can be described by a full conjoint measurement model.
Original languageEnglish
Place of PublicationEdinburgh
PublisherHeriot-Watt University
Number of pages174
Publication statusPublished - May 2008


  • Roughness perception
  • Roughness
  • Appearance
  • Models
  • 3D
  • Texture analysis
  • Texture perception
  • Surface
  • Optimisation


Dive into the research topics of 'Mathematical Models for Perceived Roughness of Three-Dimensional Surface Textures'. Together they form a unique fingerprint.

Cite this