The DynTex Database

A new issue in texture analysis is its extension to the temporal domain, a field known as Dynamic Texture analysis. Dynamic, or temporal, texture is a spatially repetitive, time-varying visual pattern that forms an image sequence with a certain temporal stationarity. In Dynamic Texture, the notion of self-similarity central to conventional image textures is extended to the spatiotemporal domain.

Dynamic textures are typically result from processes such as of waves, smoke, fire, a flag blowing in the wind, a moving escalator, or a walking crowd. Many real-world textures occurring in video databases are dynamic and retrieval should be based on both their dynamic and static features. Important tasks are thus the detection, segmentation and perceptual characterization of dynamic textures.

The ultimate goal is to be able to support video queries based on the recognition of the actual natural and artificial dynamic texture processes.

Dynamic texture research is still young, and until now no standard databases were available as in static texture analysis (e.g. the Brodatz album).

For dynamic textures, one of the few available sets to date and an interesting set to mention, is a dataset available at MIT composed of 15 black and white segmented sequences of about 170x115x120 dimension. However, this collection has a number of drawbacks: the resolution is quite low (especially temporally); there is only a single occurrence per class, and not enough classes are available for practical classification purposes; also, some of the sequences show undesirable camera motion.