Lattice Models for Context-Driven Regularization in Motion Perception

Abstract

Real-world motion field patterns contain intrinsic statistic properties that allow to define Gestalts as groups of pixels sharing the same motion property. By checking the presence of such Gestalts in optic flow fields we can make their interpretation more confident. We propose a context-sensitive recurrent filter capable of evidencing motion Gestalts corresponding to 1st-order elementary flow components (EFCs). A Gestalt emerges from a noisy flow as a solution of an iterative process of spatially interacting nodes that correlates the properties of the visual context with that of a structural model of the Gestalt. By proper specification of the interconnection scheme, the approach can be straightforwardly extended to model any type of multimodal spatio-temporal relationships (i.e., multimodal spatiotemporal context).