Ing. Jiří Filip
Supported by (ID): GAP103/11/0335
Grantor: Czech Science Foundation
Type of project: theoretical, applicational
Duration: 2011 - 2013
Publications at UTIA: list
Abstract:Real-world comprises thousands of materials having different surface textures. The appearance of these textures is a key in our everyday judgements of material properties. These judgements are based on our past experience, recognition, and usage of these materials. If we want to automate these tasks, we have to find a reliable and efficient way of digital measurement and representation of real-world materials appearance, which preserves all important visual features of genuine real-world materials.
The material visual appearance can be caused by its spectral and structural properties or their combinations.
Our previous research revealed that human perception of materials significantly depends on a type of material being observed, and that this can enhance measured data compression performance. Material appearance and its human perception are extraordinarily dependent on illumination and viewing conditions. In this project we are going to build on our previous research and develop novel methods of effective measurement of material appearance dependent on illumination and view directions. Contrary to previous measurement methods that sample these directions uniformly, and thus are extremely time-demanding and produce massive datasets, we will sample only visually significant subspace of illumination and view directions. These subspaces will differ between materials and we believe that we will be able to identify basic sampling rules for individual types of materials. We expect that novel methods of material appearance measurement will help widely access this accurate and unique technology to many applied research and industrial sectors (architecture, automotive industry, medicine, etc.).
Responsible for information: RO Last modification: