Cite as:
Tlapale É, Masson G S, Viéville T, Kornprobst P, 2007, "Model of motion field diffusion controlled by form cues" Perception 36 ECVP Abstract Supplement

Model of motion field diffusion controlled by form cues

É Tlapale, G S Masson, T Viéville, P Kornprobst

We present a model of motion integration and segmentation inspired by the primate visual system via coupled partial differential equations. These two key elements are combined in order to solve classical aperture problem and also to estimate velocity maps on complex scenes such as transparent motion or motion occlusion. Local motion estimation is solved through a recurrent V1 - MT process (Bayerl and Neumann, 2007 International Journal of Computer Vision 72 27 - 45). However, feedback diffusion is controlled via a second recurrent network (V1 - V2 like) that estimates local form information on a different spatial scale. We investigate how this architecture can reveal some integrative centre - surround MT cell properties (Huang et al, 2007 Neuron 53 761 - 770). We evaluate our model on several synthetic and real sequences against both neuronal and psychophysical data on the dynamics of motion integration.

[Supported in part by the EC IP project FP6-015879, FACETS.]

These web-based abstracts are provided for ease of seaching and access, but certain aspects (such as as mathematics) may not appear in their optimum form. For the final published version of this abstract, please see
ECVP 2007 Abstract Supplement (complete) size: 2091 Kb