THE DETECTION OF STRUCTURE IN GLASS PATTERNS - PSYCHOPHYSICS AND COMPUTATIONAL MODELS

Experiments are reported which examine the judgement of the mean orien tation of textures composed either of short lines or dipoles (Glass pa tterns). The effects of element length, density, and orientation varia tion are described, Psychophysical data are compared with predictions from four schemes for extracting features from Glass patterns: token m atching, isotropic filtering, oriented filtering, and ''adaptive'' fil tering (selection of local peak output from multiply oriented filters) , Glass patterns are spatially broadband but only contain orientation structure at a narrow range of scales making them suitable for examini ng how filter size is selected for texture processing, A criterion for scale selection is proposed: that local variation of feature orientat ion should be minimized, Simulations indicate that neither models usin g isotropic filtering nor token matching achieve human levels of perfo rmance on certain tasks, Adaptive filtering, operating at a scale sele cted using the criterion described, provides good agreement with the p sychophysical data reported and is a practical scheme for deriving fea tures using oriented filters. (C) 1997 Elsevier Science Ltd.