Part of monograph

Vácha P., Haindl M.

Adam Herout . ,

Pattern Recognition, Recent Advances

In-Teh

Vukovar, Croatia

2010

978-953-7619-90-9

Mgr. Pavel Vácha, Ph.D. (ÚTIA - Oddělení rozpoznávání obrazů)

illumination invariants, textural features, Markov random fi

Content-based image retrieval systems (CBIR) typically query large image databases based on some automatically generated colour and textural features. Optimal robust features should be geometry and illumination invariant. Although image retrieval has been an active research area for many years this difficult problem is still far from being solved. We introduce fast and robust textural features that allow retrieving images with similar scenes comprising colour textured objects viewed with different illumination. The proposed textural features that are invariant to illumination spectrum and extremely robust to illumination direction. They require only a single training image per texture and no knowledge of illumination direction, brightness or spectrum. These feature utilises utilise illumination invariant features extracted from three different Markov random field (MRF) based texture representations. The proposed illumination insensitive features are compared favourably with the most frequently used features like the Local Binary Patterns, steerable pyramid and Gabor textural features, respectively. The superiority of our proposed invariant features is demonstrated on two different texture databases. The first one is Outex texture database, which consists in hundreds of textures illuminated with three different illumination spectra. The second one, BTF database of Bonn University, comprises in the most advanced representation for realistic real-world materials - Bidirectional Texture Function (BTF) textures.