INTRINSIC 2-DIMENSIONAL FEATURES AS TEXTONS

We suggest that intrinsic two-dimensional (i2D) features, computationa lly defined as the outputs of nonlinear operators that model the activ ity of end-stopped neurons, play a role in preattentive texture discri mination. We first show that for discriminable textures with identical power spectra the predictions of traditional models depend on the typ e of nonlinearity and fail for energy measures. We then argue that the concept of intrinsic dimensionality, and the existence of end-stopped neurons, can help us to understand the role of the nonlinearities. Fu rthermore, we show examples in which models without strong i2D selecti vity fail to predict the correct ranking order of perceptual segregati on. Our arguments regarding the importance of i2D features resemble th e arguments of Julesz and co-workers regarding textons such as termina tors and crossings. However, we provide a computational framework: tha t identifies textons with the outputs of nonlinear operators that are selective to i2D features. (C) 1998 Optical Society of America.