We propose to extend a state of the art bio-inspired model for optic flow computation through adaptive processing by focusing on the role of local context indicative of the local velocity estimates reliability. We set a network structure representative of cortical areas V1, V2 and MT, and incorporate three functional principles observed in primate visual system: contrast adaptation, adaptive afferent pooling and MT diffusion that are adaptive dependent upon the 2D image structure (Adaptive Motion Pooling and Diffusion, AMPD). We assess the AMPD performance on Middlebury optical flow estimation dataset, showing that the proposed AMPD model performs better than the baseline one and its overall performance is comparable with many computer vision methods.
Adaptive Motion Pooling and Diffusion for Optical Flow Computation
Manuela Chessa;Fabio Solari
2017-01-01
Abstract
We propose to extend a state of the art bio-inspired model for optic flow computation through adaptive processing by focusing on the role of local context indicative of the local velocity estimates reliability. We set a network structure representative of cortical areas V1, V2 and MT, and incorporate three functional principles observed in primate visual system: contrast adaptation, adaptive afferent pooling and MT diffusion that are adaptive dependent upon the 2D image structure (Adaptive Motion Pooling and Diffusion, AMPD). We assess the AMPD performance on Middlebury optical flow estimation dataset, showing that the proposed AMPD model performs better than the baseline one and its overall performance is comparable with many computer vision methods.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.